Post-Graduate Research Report A multicentre study to evaluate an in-house multiparameter immunophenotypic panel to identify precursor B-cells in the determination of measurable residual disease in paediatric B-cell acute lymphoblastic leukaemia MMed candidate : Dr Zanré Nell MBBCh Wits, 449 647 MMed(Haem)(Wits) Supervisor 1 : Professor Deborah Glencross MBBCh(Wits), MMed(Haem)(Wits) HOD: CD4, Leukaemia and HIV Immunology Laboratory, CMJAH Director: CD4 National Priority Programme, NHLS Supervisor 2 : Professor Jennifer Geel MBBCh(UCT), FC Paed(SA), Cert Med Onc(Paed), MMed (Paed) HOD: Paediatric Oncology, CMJAH and Wits DGMC Lecturer, Wits (joint appointment) TABLE OF CONTENTS List of tables and figures .................................................................................................................................. i – ii Abbreviations ................................................................................................................................................. iii – iv Definitions and terminology .......................................................................................................................... v – vii Abstract ....................................................................................................................................................... viii – ix 1. Background and introduction ................................................................................................................... 1 – 7 1.1 Overview of B-cell acute lymphoblastic leukaemia ......................................................................... 1 1.2 Measurable residual disease in haematological malignancies ................................................... 1 – 2 1.3 Overview of existing B-cell ALL MRD strategies .................................................................... 3 – 5 1.4 Role of flow cytometry in B-cell ALL MRD ............................................................................ 6 – 7 2. Aims and objectives .................................................................................................................................... 7 – 8 2.1 Study aim .......................................................................................................................................... 7 2.2 Study objectives ............................................................................................................................... 8 3. Materials and methods ............................................................................................................................. 8 – 15 3.1 Ethical considerations ...................................................................................................................... 8 3.2 Study design ............................................................................................................................... 8 – 9 3.3 Sample size and eligibility ....................................................................................................... 9 – 10 3.4 Research laboratory procedures ............................................................................................. 10 – 14 3.4.1 KaluzaCTM in-house haematogone data analysis protocol development and technicalities ................................................................................................. 10 – 14 3.4.2 Study-specific bench procedure .......................................................................... 14 3.5 Data collection and patient records ................................................................................................ 15 4. Data analysis ........................................................................................................................................... 15 – 19 4.1 Part 1 ........................................................................................................................................ 15 – 16 4.2 Part 2 ..................................................................................................................................... 16 – 19 5. Results ...................................................................................................................................................... 19 – 24 5.1 Part 1 ...................................................................................................................................................... 19 5.2 Part 2 ............................................................................................................................................ 19 – 24 5.2.1 Overview of eligible patients ....................................................................... 19 – 20 5.2.2 Qualitative method analysis ......................................................................... 20 – 21 5.2.3 Other noteworthy findings and observations ............................................... 21 – 24 5.3 Conclusive remark ............................................................................................................................... 24 6. Discussion ................................................................................................................................................ 24 – 27 7. Acknowledgements ........................................................................................................................................ 27 8. Research funding ............................................................................................................................................. 27 9. Ethics and other approval requirements ....................................................................................................... 28 10. References ................................................................................................................................................. 28 – 34 11. Appendices ............................................................................................................................................... 35 – 75 Appendix A Table 1.4.1. Normal maturation of the B-cell .............................................................. 35 Appendix B Table 1.4.2. Useful markers to differentiate between LAIP and haematogone populations within B-cell ALL MRD bone marrow samples .......... 36 – 37 Appendix C Figure 3.4.1.1. In-house haematogone data analysis protocol using the Beckman Coulter KaluzaCTM software system ............................................................ 38 – 40 Appendix D Figure 3.4.1.4. Example of the immunophenotypic analysis of a normal paediatric bone marrow aspirate sample using the in-house haematogone data analysis protocol ............................................................................................ 41 – 43 Appendix E Table 3.5.1. Example of a single patient’s diagnostic and follow-up captured information .............................................................................................. 44 Appendix F Table 4.1.1. Documentation of verification findings of the manufacturer-specified performances of reagents CD58-FITC and CD81-APC-H7 by their addition to remnant normal paediatric BMA samples to identify precursor B-cells .............................................................................................. 45 – 48 Appendix G Figure 4.1.1. Carry-over assessment of the Beckman Coulter Navios flow cytometer ........................................................................................... 49 – 51 Appendix H Table 4.1.2. Individual lower limit of quantification and precision of the ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels ............ 52 – 53 Appendix I Figure 4.2.2. Example of the full immunophenotypic analysis of a patient’s diagnostic remnant BMA sample using the in-house haematogone data analysis protocol ............................................................................................ 54 – 56 Appendix J Figure 4.2.4. Example of the full immunophenotypic analysis of a patient’s follow-up remnant BMA sample using the in-house haematogone data analysis protocol, showing significant residual disease ................................. 57 – 59 Figure 4.2.5. Example of the full immunophenotypic analysis of a patient’s follow-up remnant BMA sample using the in-house haematogone data analysis protocol, showing no overt immunophenotypic evidence of residual disease (i.e. no distinct clustering of LAIP) .................................................... 60 – 62 Appendix K Table 4.2.1. Raw data assessing the association between categorical (binary) outcome and independent exposure variables from which logistic regression analysis was performed .................................................................... 63 – 65 Table 4.2.2. Tables of test by response: dependent binary outcome variable versus independent exposure variables .................................................................. 66 Appendix L Ethics and research approval documents .................................................................... 67 – 72 Wits HREC (Medical) supervisor clearance certificate (M1704129) ................... 67 Wits HREC (Medical) MMed candidate clearance certificate (M220153) .. 68 – 69 NHLS AARMs: Permission to use NHLS facilities (PR2222608) ....................... 70 Permission to use the Beckman Coulter KaluzaCTM software system ................. 71 CMJAH CEO approval to conduct research ......................................................... 72 Appendix M Plagiarism documentation .......................................................................................... 73 – 75 i LIST OF TABLES AND FIGURES Table 1.4.1. Normal maturation of the B-cell ............................................................................................ 35 Table 1.4.2. Useful markers to differentiate between LAIP and haematogone populations within B-cell ALL MRD bone marrow samples .......................................................................... 36 – 37 Table 3.3.1. Sample size estimation .......................................................................................................... 10 Table 3.3.2. Research study patient inclusion and exclusion criteria .......................................................... 10 Table 3.5.1. Example of a single patient’s diagnostic and follow-up captured information ....................... 44 Table 4.1.1. Documentation of verification findings of the manufacturer-specified performances of reagents CD58-FITC and CD81-APC-H7 by their addition to remnant normal paediatric BMA samples to identify precursor B-cells .................................................. 45 – 48 Table 4.1.2. Individual lower limit of quantification and precision of the ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels ...................................................... 52 – 53 Table 4.2.1. Raw data assessing the association between categorical (binary) outcome and independent exposure variables from which logistic regression analysis was performed ................... 63 - 65 Table 4.2.2. Tables of test by response: dependent binary outcome variable versus independent exposure variables ................................................................................................................. 66 Table 5.2.2.1. Sensitivity and specificity measures for ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels ................................................................................................... 21 Figure 3.2.1. Components of the in-house immunophenotypic panel, including primary backbone (CD45-KrO), secondary backbone (CD19-PC7), tertiary (CD10-APC700; CD34-APC and other mAbs) and assessed markers of interest (CD58-FITC and CD81-APC-H7) .................. 9 Figure 3.4.1.1. In-house haematogone data analysis protocol using the Beckman Coulter KaluzaCTM software system ......................................................................................................... 38 – 40 Figure 3.4.1.2. Normal paediatric bone marrow aspirate leucocyte populations, including maturing precursor B-cells (haematogones) represented by varying shades of orange .................. 12 ii Figure 3.4.1.3. Boolean gating strategy ..................................................................................................... 12 Figure 3.4.1.4. Example of the immunophenotypic analysis of a normal paediatric bone marrow aspirate sample using the in-house haematogone data analysis protocol ............... 41 – 43 Figure 4.1.1. Carry-over assessment of the Beckman Coulter Navios flow cytometer .................. 49 – 51 Figure 4.2.1. Example of a paediatric B-cell ALL with strategies employed to identify the LAIP ........... 17 Figure 4.2.2. Example of the full immunophenotypic analysis of a patient’s diagnostic remnant BMA sample using the in-house haematogone data analysis protocol ............................. 54 – 56 Figure 4.2.3. Example of a paediatric B-cell ALL residual disease with distinct clustering noted in the dual plots incorporating the discriminatory markers of interest in the in-house haematogone data analysis protocol ................................................................................. 18 Figure 4.2.4. Example of the full immunophenotypic analysis of a patient’s follow-up remnant BMA sample using the in-house haematogone data analysis protocol, showing significant residual disease .......................................................................................................... 57 – 59 Figure 4.2.5. Example of the full immunophenotypic analysis of a patient’s follow-up remnant BMA sample using the in-house haematogone data analysis protocol, showing no overt immunophenotypic evidence of residual disease (i.e. no distinct clustering of LAIP) 60 – 62 Figure 5.2.1.1. Research study patient and sample eligibility .................................................................... 20 Figure 5.2.2.1. Receiver Operating Characteristic curve used to assess the association between the binary outcome (cPCR) and independent exposure variables (ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels) ......................................................... 21 iii ABBREVIATIONS  ALB Abnormal B-progenitor cells  ALL Acute lymphoblastic leukaemia  APC-H7 Allophycocyanin-cyanine dye analogue  AUC Area under the curve  BFM Berlin-Frankfurt-Münster (BFM)  BMA Bone marrow aspirate  CD Cluster of differentiation  CHBAH Chris Hani Baragwanath Hospital  CMJAH Charlotte Maxeke Johannesburg Academic Hospital  DGMC Donald Gordon Medical Centre  DNA Deoxyribonucleic acid  ECD Phycoerythrin-Texas Red conjugate  EQA External quality assurance  FcR Fragment region receptor  FISH Fluorescence in situ hybridisation  FITC Fluorescein isothiocyanate  FS Forward scatter  HSCT Haemopoietic stem cell transplant  IgH Immunoglobulin heavy chain  IQC Internal quality control  LAIP Leukaemia-associated immunophenotype  LDT Laboratory-developed test  LLOQ Lowest limit of quantification  LMIC Low and middle income countries  LOD Limit of detection  mAb Monoclonal antibody  MFC Multiparameter flow cytometry iv  MFI Mean fluorescence intensity  MRD Measurable residual disease  M2 Myeloid-cell 2  NSB Nonspecific binding  NHLS National Health Laboratory Service  PBS Phosphate-buffered saline  PCR Polymerase chain reaction • cPCR (conventional PCR) • RQ-PCR (real-time quantitative PCR)  ROC curve Receiver Operating Characteristic curve  SOP Standard operating procedure  SS Side scatter v DEFINITIONS AND TERMINOLOGY  Boolean gating Also logic gating Refers to individualised flow cytometric gating strategies allowing for the accurate identification of leucocyte populations based on an “AND”, “OR” and “NOT” logic  Comparability Also reliability, agreement and accuracy Refers to the extent of agreement between results generated by a method and its true value  Carry-over assessment Refers to the analysis for any discrete amount of analyte carried by a measurement system from one sample reaction to the next Involves the calculation of the Broughton formula and T-test  Compensation Process by which intrinsic spectral overlap between fluorochromes is mathematically eliminated using computer software  EQA External quality assurance Refers to the planned and systematic set of quality activities focused on providing confidence that quality requirements will be fulfilled  Fluorochrome Fluorescent molecules attached to nucleotides that absorb energy from light and raise to an excited state where they emit energy as fluoresce when they return to ground state  FRET Fluorescence resonance energy transfer Refers to a visual signal technique detecting fluorochrome-emitted energy that is transferred from one fluorochrome to the next  Haemodilution Refers to the dilution of a bone marrow aspirate sample with sequential pulls on sample acquisition resulting in a sample resembling peripheral blood constituents vi  IQC Internal quality control Refers to a set of activities or techniques whose purpose is to ensure that all quality requirements are being met  LDT Laboratory-developed test Refers to an assay that has been developed in a single laboratory, or an approved test system which has been modified or used outside its intended scope  LOD Limit of detection Refers to the lowest quantity of a component that can reliably be detected with a given analytical method with stated confidence or statistical significance  Logistic regression A statistical method commonly used for the analysis of binary outcome variables when exposed to single or multiple exposure variables.  LLOQ Lowest limit of quantification Refers to the lowest amount of an analyte detected by a test method with stated probability provided accuracy and precision have been demonstrated  MFI Mean fluorescence intensity Measurement used to define the mean intensity and level of mAb expression, directly proportional to instrument optimisation (IQC)  NPV Negative predictive value Refers to the proportion of test positives that are truly positive  PPV Positive predictive value Refers to the proportion of test negatives that are truly negative  Precision Refers to the closeness of agreement between repeated measurements vii in a sample  Sensitivity Refers to the proportion of true positives correctly identified as such  Specificity Refers to the proportion of true negatives correctly identified as such  SOPs Standard operating procedures Refers to a set of written instructions that describe in detail how a laboratory process is to be carried out so as to ensure standardisation within the laboratory  Spectral overlap Refers to the leakage of donor dye fluorescence and acceptor dye diminished fluorescence as a result of the introduction of multiple fluorochromes capable of being excited by a single light source  Target population Refers to leukaemic cells of interest that express a particular leukaemia-associated immunophenotype (LAIP)  Validation A defined process that provides objective evidence that a test system meets the requirements for intended use by manufacturer- specified claims. Minimum requirements include accuracy, precision and linearity assessment.  Verification An abbreviated process to demonstrate that a test system performs in substantial compliance to previously established manufacturer claims viii ABSTRACT Background: Periodic assessment of measurable residual disease (MRD) is an important prognostic factor in the management of paediatric B-cell acute lymphoblastic leukaemia (ALL). Conventional polymerase chain reaction (cPCR) and multiparameter flow cytometry (MFC) are well-established in MRD determination, the latter with no current optimal immunophenotypic panel by international consensus. Objective: To determine whether an in-house immunophenotypic panel containing the discriminatory CD58-FITC (cluster of differentiation; fluorescein isothiocyanate) marker compares with cPCR in the detection of paediatric B-cell ALL MRD. Methods: This prospective descriptive validation study was performed on diagnostic and follow-up bone marrow aspirate samples, comparing an in-house immunophenotypic panel against the standardised commercial ClearLLab 10CTM B-cell/myeloid cell-2 (M2) panels in MRD assessment. These findings were then compared to cPCR to determine individual panel performance and predictive power. Results: Both immunophenotypic panels demonstrated 100% concordance in the identification of the leukaemia-associated immunophenotype (LAIP) on all diagnostic samples. The in-house immunophenotypic panel showed a higher sensitivity and specificity, and greater association with cPCR in MRD assessment in follow-up samples. In combination with shared backbone markers of the ClearLLab 10CTM B-cell/M2 panels, inclusion of CD58-FITC and CD81-APC-H7 (allophycocyanin- cyanine dye) proved most informative in accurate distinction between regenerating B-cell precursors and residual leukaemic cells. Conclusion: This work confirms the findings of previous studies, where discriminatory marker CD58- FITC in combination with backbone informative markers demonstrates both superior diagnostic and monitoring utility in paediatric B-cell ALL. The in-house immunophenotypic panel offers an attractive, comparable alternative in MRD determination in this patient population whilst awaiting cPCR results, ix raising the possibility of earlier clinical decision-making with potential improvement of morbidity and mortality outcomes. Key words: B-cell acute lymphoblastic leukaemia; measurable residual disease; conventional polymerase chain reaction; multiparameter flow cytometry. 1 1. BACKGROUND AND INTRODUCTION 1.1. Overview of B-cell acute lymphoblastic leukaemia B-cell acute lymphoblastic leukaemia (ALL) refers to the malignant transformation and proliferation of B- cell lymphoid progenitor cells arising within the bone marrow.(1) A classical bimodal age distribution has been described, with the majority of cases within the paediatric population, and a second peak later in adult life.(2,3) Accurate epidemiological data reflecting the true incidence of B-cell ALL within the sub-Saharan African paediatric population is limited by resource constraints, including accessibility to healthcare and cancer surveillance strategies employed.(3–5) High rates of loss-to-follow-up and delay in intervention compound the untimely loss of life in low and middle income countries (LMIC) in particular.(4) Risk stratification and prognostication is central to paediatric B-cell ALL management, allowing for appropriate initial regimen choice, individualised treatment consideration and eligibility for haemopoietic stem cell transplantation (HSCT).(2,6) Traditional prognostic factors include patient age at diagnosis, baseline white cell count and early response to treatment by time to remission and measurable residual disease (MRD) determination, with recent molecular technologies having identified defined genetic abnormalities that contribute to this risk stratification strategy.(2,6–8) The majority of newly-diagnosed paediatric patients attain both morphological and molecular remission after completion of the induction phase of treatment with conventional cytotoxic chemotherapy regimens.(9) 1.2. Measurable residual disease in haematological malignancies MRD refers to the subclinical presence of residual leukaemic cells that are below the level of morphological detection (5% cell threshold of detection) using molecular tools such as flow cytometry, cytogenetics and polymerase chain reaction (PCR)-based approaches.(9–12) The prognostic MRD threshold (also limit of detection, LOD) of ≤ 0.01% suggests that, should a patient have cellular MRD levels above this demarcated value at a specified time point during therapy, it is associated with a significantly higher risk of future relapse.(13–15) The definition of true MRD negativity is debatable and depends on the MRD assay and corresponding sensitivity for detection of very small populations of residual leukaemic cells.(10) 2 A negative MRD status implies that no MRD is detected with a high level of certainty, using molecular techniques that have been validated to measure low quantitative levels of residual leukaemic cells.(10,16) In paediatric B-cell ALL, MRD status informs the clinician about remaining disease burden after induction therapy using institution-specific, modified Berlin-Frankfurt-Münster (BFM) treatment protocols, reflecting chemosensitivity and treatment response.(15,17) Furthermore, MRD status forms a major predictor of overall and disease-free survival, with the potential for earlier clinical decision-making pertaining to regimen modification (i.e. therapy minimisation versus intensification, or review of HSCT eligibility), improving morbidity and mortality in this patient population.(2,6,9,11,18) The level of MRD in paediatric patients pre-HSCT conditioning has a significant impact on post-transplant outcomes and prediction for post-transplant disease relapse.(11) B-cell ALL MRD negativity has also been used as a surrogate therapeutic endpoint for drug approval in clinical trials over the last two decades.(16) According to a number of studies, periodic B-cell ALL MRD measurements at days 15, 33 and 78 have proven to provide information for the identification of very early treatment responders as well as a small group of poor treatment responders. (9,10,19) MRD measurements at early and later treatment time points correlates with long-term relapse-free survival.(15) However, despite MRD exclusion patients may still relapse due to underlying clonal evolution and/or persistence of, and subsequent clonal evolution of subclonal populations with an “antigenic shift” as these cells acquire new immunophenotypes.(9,10,13,20) MRD determination requires skilled interpretation to differentiate between normal haemopoietic cells (including B-cell lineage precursors or haematogones) and aberrant leukaemic target populations.(19,21,22) Expanded populations of haematogones are observed in children, comprising up to 50% of marrow cellularity and as a result, pose a diagnostic challenge in paediatric B-cell ALL MRD analysis. 3 1.3. Overview of existing B-cell ALL MRD strategies Classical MRD strategies include conventional PCR (cPCR) for the identification of B-cell immunoglobulin heavy chain (IgH) gene re-arrangements, real time quantitative PCR (RQ-PCR) for defined genetic abnormality detection (if present), flow cytometry to identify the leukaemia-associated immunophenotype (LAIP), and the detection of specific genetic aberrations by cytogenetics [chromosomal analysis and fluorescence in situ hybridisation (FISH)].(7,10,18,23,24) For the purpose of this research report, further discussion is focused around flow cytometry and cPCR, both methods demonstrating good correlation in MRD detection in paediatric B-cell ALL.(7,9,25) The cPCR method for B-cell ALL MRD determination is well-standardised, being the most studied molecular technique for MRD detection, large scale validations having been conducted worldwide using various manufacturer-specified kits.(8,25,26) The general principle involves rapid in vitro enzymatic amplification of specific deoxyribonucleic acid (DNA) segments using sequence-specific single-stranded primers that target desired DNA sequences within conserved genetic framework regions. Endpoint analysis of amplified products may then be interpreted using various assays, including fragment analysis by capillary electrophoresis.(8,25) The National Health Laboratory Service (NHLS) at Charlotte Maxeke Johannesburg Academic Hospital (CMJAH) currently uses the IdentiClone IgH Gene Clonality PCR kit for B-cell IgH gene re-arrangement detection, with a sensitivity of 93% and specificity of 92% when compared to Southern Blot analysis.(8,26) According to various authors, the LOD of cPCR ranges between 0.01 – 0.001%, which may be increased using nested PCR or higher DNA input strategies, the latter not currently employed in our laboratory.(8,25) In comparison to its superior RQ-PCR counterpart where a specific defined genetic abnormality has been identified, cPCR offers a major advantage where it may be applied to the majority of B-cell ALL patients, with the exception of a few cases where IgH gene re-arrangement cannot be identified at diagnosis (i.e. polyclonal or failed initial analysis).(25) 4 However, despite the accurate and sensitive detection of low frequencies of B-cell ALL leukaemic cells by amplification of small DNA quantities, cPCR has its limitations.(8–11,14,25) In addition to the risk of contamination and initial start-up cost, the modality does not offer the opportunity for real-time, same-day clinical decision-making regarding patient management. Furthermore, owing to patient-specific unique variable, diversity and joining gene segment re-arrangements, each case requires individualised assessment.(25) These gene re-arrangements are not essential for leukaemic cell survival and in the advent of clonal evolution, patients may relapse with clones that do not harbour the initial diagnostic gene re- arrangement, leading to possible false negative interpretation of MRD status.(9–11,25) Flow cytometry assigns an immunophenotypic profile to a target population of cells, identifying physical characteristics including cell size and complexity, as well as antigen expression. Antigen-specific fluorochrome-labelled monoclonal antibodies (mAbs) are added to the sample, excited by laser light source emission at varying wavelengths and detected by fluorescent detectors and mirrors which are in turn measured by flow cytometric software.(27,28) In comparison to cPCR, flow cytometry has a quicker turnaround time, offering timeous diagnostic reporting to facilitate appropriate patient management. Furthermore, the modality is able to detect subclonal populations and is less likely to pick up apoptotic blast cells which may yield false positive MRD results.(7) Limitations of flow cytometry include the need for expert analysis, particularly if a very small target population is present, as well the requirement for a minimum number of cellular events for accurate interpretation. Furthermore, when analysing a bone marrow aspirate (BMA) sample the quality of the sample directly implicates immunophenotypic assessment, particularly if a haemodilute specimen is obtained, as this may underestimate residual leukaemic burden and yield a false negative MRD status.(11,17,18,27–29) Multiparameter flow cytometry (MFC) measures several cellular characteristics simultaneously for total cell analysis by incorporating a larger number of novel fluorochrome-labelled mAbs to better identify a target population.(30,31) Together with advanced instrumentation and software where “rare event” analysis may be employed to artificially amplify and enable easy visualistion of very small number of leukaemic 5 cells, MFC improves the overall precision of specimen immunophenotypic analysis.(11,28,29,32) MFC for LAIP detection has a LOD similar to cPCR ranging between 0.01 – 0.001%, which is increased when more than 50 000 events are analysed.(25) Determination of the assay’s sensitivity and specificity is dependent on the number of fluorochrome-labelled mAbs detected by laser light application. In two recent multicentre observational studies, one conducted in Europe and North America and the other in South Africa, the diagnostic accuracy of the ClearLLab 10CTM panels using the Beckman Coulter’s KaluzaCTM software system was determined when applied to patient specimens suspected of a haematological malignancy. Both studies yielded high sensitivities (96% and 100%) and specificities (95% and 100%) respectively within their cohorts.(33–36) Participation in proficiency testing schemes and accreditation programmes with widespread incorporation of educational material may ensure transparency and reliability in MRD monitoring using MFC. (18,29) Compliance to manufacturer-described instrument setup additionally ensures reproducible and institution- specific standardised sample analysis.(25,33,34) Novel software tools may facilitate a more objective immunophenotypic interpretation, including specific automated gating strategies and maturation pathway analysis in an attempt to standardise the identification of various B-cell populations in MRD samples.(14,37) A major limiting factor with MFC is the concept of “spectral overlap” with the introduction of tandem conjugate (multiple) mAbs into the system.(38,39) These tandem conjugates expand the pool of mAbs that may be detected, measuring fluorescent resonance energy transfer by excitation of multiple acceptor dyes using a single laser emitting light at a particular emission spectrum. Commonly used tandem conjugates vary in their transfer capabilities from donor to acceptor dyes across mAb lots over time, resulting in the leakage of donor dye fluorescence and diminished acceptor dye fluorescence (i.e. “spectral overlap”).(39) This may be overcome by the application of “colour compensation” using electronic elimination of duplicated fluorescent emission.(38) Lastly, the antigenic shift that may occur with clonal evolution may reduce the assay’s accuracy, leading to possible false negative MRD interpretation.(17) 6 1.4. Role of flow cytometry in B-cell ALL MRD To correctly interpret MFC immunophenotypic information, a sound understanding of normal cell biology and antigenic expression (cluster of differentiation, CD) is necessary. Additionally, strict adherence to sample preparation and acquisition of as large a number of cellular events as possible is required.(40) Normal B-cell education and maturation occurs within primary and secondary lymphoid organs, with gain and/or loss of certain antigens at variable levels of expression occurring during B-cell development (see Table 1.4.1; Appendix A). (33,41,42) It is important to distinguish between haematogones and residual or subclonal leukaemic cells, as incorrect interpretation owing to immunophenotypic overlap may adversely affect patient care.(19,21,22,43) Table 1.4.2 (see Appendix B) summarises the usefulness of previously explored antigen markers in isolation and in combination with other maturity markers in the attempt to differentiate between the two populations by MFC.(12,19,20,22,33) Various research groups (including Becton Dickinson and Beckman Coulter) have attempted to develop standardised B-cell ALL MRD MFC data analysis protocols, however no one optimal panel exists that minimises cost and spillover effect, whilst maximising relevant immunophenotypic information obtained.(14,38) Panel selection may be individualised as a laboratory-developed test (LDT), dependent on resource and expertise availability, whilst ensuring retention of the predictive power of larger panels.(13,21,44,45) A wide range of surface mAbs have been validated in the determination of B-cell ALL MRD. Many of these mAbs are used regularly outside of the B-cell ALL diagnostic scope, as well as less familiar markers requiring more expert interpretation.(20) Consensus regarding the precise combination of aberrant markers which identify the presence of the LAIP and subclones, if present, has not yet been reached.(19,20) Novel markers may perform well in individual validations, but may not add value in combination with other markers, implying that their value is dependent on clinical context.(20,44) Shaver elegantly describes five important considerations when selecting an appropriate panel for B-cell ALL diagnostic and follow-up MRD determination.(20) An initial gating strategy may be employed to assign lineage to the leucocyte 7 population within the sample, followed by markers of immaturity to identify the target population. Aberrancies detected at presentation or acquired during induction therapy may include evidence of asynchronous and aberrant maturation patterns, as well as inappropriate myeloid antigens and certain markers associated with specific B-cell ALL genetic abnormalities.(11,20,33,46) Further panel design may include a targeted therapeutic marker, whereby a reduction and subsequent normal maturation pattern attained by the relevant informative marker may indicate the absence of MRD.(40) Several studies have rationalised a limited number of mAbs with comparable sensitivity to cPCR for MRD determination in B-cell ALL by employing “design-test-evaluate-redesign” strategies.(18,39) General consensus amongst these studies includes a combination of at least two informative, LAIP-defining aberrant markers in addition to basic “backbone” markers confirming lineage and immaturity of the target population. Panel optimisation involves choosing informative markers that are both discriminatory enough to detect subsets of B-cell ALL cases that comprise less striking abnormal immunophenotypes, and detect subtle differences from background developing precursor B-cells.(16,44) The combinations of (CD38/CD58), (CD58/CD81) and (CD81/CD123) are most commonly proposed using this panel design strategy, with CD38 being the most informative across studies, with superior diagnostic utility of CD58 having been proposed.(14,16,22,40,44,47) Adoption of these mAb combinations assists in the development of an in-house immunophenotypic formulation, standardisation and informative MRD predictive values.(20) At the time of submission of this research report, the predictive power of cPCR against the validated ClearLLab 10CTM panels had not yet been determined in the wider scope of laboratory practice. 2. AIMS AND OBJECTIVES 2.1. Study aim To determine whether an in-house immunophenotypic panel containing the discriminatory CD58-FITC (fluorescein isothiocyanate) marker compares with cPCR in the detection of paediatric B-cell ALL MRD. 8 2.2. Study objectives 2.2.1. To verify the manufacturer-specified performances of reagents CD58-FITC and CD81-APC-H7 (allophycocyanin-cyanine dye) using remnant paediatric bone marrow samples submitted for investigation in which no active disease or infiltrate has been identified morphologically and/or by immunophenotypic analysis. 2.2.2. To compare the in-house immunophenotypic panel against the validated ClearLLab 10CTM B- cell/myeloid cell-2 (M2) in diagnostic and follow-up bone marrow investigations of paediatric patients with B-cell ALL. 2.2.3. To evaluate the predictive power of the in-house immunophenotypic panel as compared to the cPCR technique for paediatric B-cell ALL MRD determination. 3. MATERIALS AND METHODS 3.1. Ethical considerations Ethics and research approval information and documents may be reviewed under “Ethics and research approval documents”. Individual patient consent was not required as only remnant BMA samples were used for secondary analysis. All flow cytometric data analysed was anonymised with patient identifiers removed for the purpose of this study. 3.2. Study design This was a prospective multicentre study conducted during the period of 01 June 2021 – 31 December 2022. BMA samples were received from specified paediatric facilities at the national referral flow cytometry laboratory at the CMJAH NHLS. Pilot work conducted prior to ethics clearance was covered under the principal supervisor’s ethics clearance for the evaluation and validation of kits, reagents, data analysis protocols and methods. As part of research and development, Professor Deborah Glencross initiated investigation into the addition of drop-in CD58-FITC and CD81-APC-H7 as an alternative to the validated ClearLLab 10CTM B-cell/M2 panels in the identification of haematogones and leukaemic cells 9 in BMA samples of paediatric patients with B-cell ALL. The previously mentioned “design-test- evaluate-redesign” strategy was employed for this component of the study. This qualitative validation study was conducted in two parts: Part 1 comprised verification of the manufacturer-specified performance of the reagents included in the in-house immunophenotypic panel (see Figure 3.2.1) against the validated ClearLLab 10CTM panels in the identification of precursor B-cells in paediatric BMA samples with no demonstrable disease. During this process, an in-house haematogone data analysis protocol was developed using the validated Beckman Coulter KaluzaCTM software system. The ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels with their respective data analysis protocols were then analysed in Part 2, using both diagnostic and follow-up paediatric B-cell ALL BMAs for final determination of predictive power as compared to cPCR in MRD assessment. CD58-FITC CD38-PE CD123-ECD CD3-PC5.5 CD19-PC7 CD34-APC CD10-APC- A700 CD81-APC- H7 CD20-PB CD45-KrO Figure 3.2.1. Components of the in-house immunophenotypic panel, including primary backbone (CD45-KrO), secondary backbone (CD19-PC7), tertiary (CD10-APC700; CD34-APC and other mAbs) and assessed markers of interest (CD58-FITC and CD81-APC-H7). CD, cluster of differentiation; FITC, fluorescein isothiocyanate; APC- H7, allophycocyanin-cyanine dye. Abbreviations of the non-assessed markers are not further described. 3.3. Sample size and eligibility In their observational study to assess the genetic landscape of B-cell ALL in the South African Johannesburg state sector setting, Vaughn and colleagues identified all newly-diagnosed cases captured on the CMJAH NHLS flow cytometry database during a 36-month period between 2016 and 2019.(49) Their data showed that 108/461 cases (23.4%) comprised a B-cell immunophenotype, of which 82 cases (75.9%) fell within the paediatric age group (≤ 18 years old); approximately 36 – 40 new cases identified annually within the hospital complexes referring their molecular testing to CMJAH.(48) The sample size (n=45) for this study was determined using the statistical method of sensitivity and specificity (see Table 3.3.1).(50,51) 10 Table 3.3.1. Sample size estimation Width of 95% CI SN SP Disease prevalence Sample size for SN Sample size for SP N 0.14 0.9 0.85 0.4 45 42 45 * CI, confidence interval; SN, sensitivity; SP, specificity; N, total number of samples. The following inclusion and exclusion criteria were strictly adhered to in sample collection and eligibility determination for the duration of the study period: Table 3.3.2. Research study patient inclusion and exclusion criteria Inclusion criteria Exclusion criteria 1. Paediatric age group (<18 years old or institution-dependent paediatric age range) 2. Referral paediatric centres included CMJAH, CHBAH and DGMC 3. Patients with newly-diagnosed or relapsed paediatric B-cell ALL cases with at least one follow up BMA sample submitted during the period of 01 June 2021 – 31 December 2022 1. Paediatric acute leukaemias with a mixed immunophenotype or lineage designation not possible at diagnosis 2. Inappropriate samples on which diagnostic immunophenotyping specific for this research cannot be performed (e.g. diagnostic trephine imprints and other fluids) 3. Diagnostic or relapsed paediatric B-cell ALL BMA samples in which no addition of the in-house immunophenotypic panel and/or cPCR was performed 4. Poor quality BMA samples (e.g. clotted sample or poor cell viability) 5. Incomplete sample collection (e.g. No follow-up sample received; insufficient remnant BMA sample; diagnostic sample analysed at a different facility) ∗ CMJAH, Charlotte Maxeke Johannesburg Academic Hospital; CHBAH, Chris Hani Baragwanath Academic Hospital; DGMC, Donald Gordon Medical Centre; ALL, acute lymphoblastic leukaemia; BMA, bone marrow aspirate; cPCR, conventional polymerase chain reaction. 3.4. Research laboratory procedures 3.4.1. KaluzaCTM in-house haematogone data analysis protocol development and technicalities Under the guidance of the principle supervisor, an in-house haematogone data analysis protocol was developed using the Beckman Coulter KaluzaCTM software system, developed using the 20 verification samples collected in Part 1 of the study. The newly-devised data analysis protocol was applied to each verification sample (normal paediatric BMA samples containing precursor B-cells with no active disease 11 or infiltrate demonstrable) and tailored so that the final in-house haematogone data analysis protocol would correctly identify any precursor B-cells in the sample based on their expected immunophenotypic expression profile (see Figure 3.4.1.1; Appendix C). Several factors were taken into consideration in developing this in-house haematogone data analysis protocol, summarised below.(33,49,50) Pre-analytical data analysis considerations • Time versus CD45 dual plot: This ungated plot was included to show all events collected in sequence, evaluating fluidic perturbation during sample acquisition. Acceptable samples yield a uniform pattern of events. • Forward scatter (FS) peak and FS INT (integral) dual plot: This plot was included to exclude all doublet events (i.e. non-singly occurring cells), debris and dead or dying cells. If included in analysis, these events would yield marker overexpression and false positive interpretation. • FL3 (fluorescence; CD123) versus FL5 (CD19) dual plot: This plot allowed for the exclusion of any nonspecific binding (NSB) of mAbs to leucocyte fragment region receptors (FcRs) which may yield erroneous background fluorescence. • CD45 versus side scatter (SS) dual plot: This plot allowed for the identification of normal background leucocyte populations in the context of CD45 expression and their respective cell complexity. Haematogones were identified based on their characteristic non-discrete, beaded-like appearance. Haematogones may express very dim CD45 and should not accidentally be excluded from analysis. • Colour compensation: Intrinsic spectral overlap was eliminated in order to avoid artificial population identification and false positive interpretation. SS versus CD dual plots Using a CD45-gating strategy, separation and identification of the various leucocyte populations present was done using SS versus CD plots. In addition to the haematogone populations within normal paediatric 12 samples, other discernable populations identified in these plots included myeloblasts, monocytes, basophils, granulocytes, T-cells and background mature B-cells (see Figure 3.4.1.2). Figure 3.4.1.2. Normal paediatric bone marrow aspirate leucocyte populations, including maturing precursor B-cells (haematogones) represented by varying shades of orange. CD, cluster of differentiation; SS, side scatter; ALL, acute lymphoblastic leukaemia. Boolean gating strategy employed The following Boolean gating strategy was employed to correctly identify the precursor B-cell population within normal paediatric BMA samples, labelled “Precursor B-cells” in the in-house haematogone data analysis protocol: Figure 3.4.1.3. Boolean gating strategy. Gr, granulocytes; Mo, monocytes; NSB, nonspecific binding. In its formulation, the various leucocyte populations present were identified by applying stepwise, logical inferences and gating strategies based on the knowledge of normal cell biology and antigenic expression.(40,49) • Granulocytes were first identified on the CD45 versus SS dual plot, showing a pattern of appropriate variability in size and complexity seen in its maturation series (labelled Gr; lighter (CD45+ AND B-cells) AND (NOT Gr) AND (NOT CD3+) AND (NOT Mo) AND (NOT NSB) AND (NOT Eosinophils) 13 blue). This was followed by the identification of more mature, CD10-expressing forms on the CD10 versus SS dual plot (labelled A; dark blue), concurring with their brightest CD45 expression on the CD45 versus SS dual plot. • Monocytes were then identified on the CD45 versus SS dual plot, showing bright CD45 expression with higher side scatter (labelled Mo; green) when compared to lymphocytes. • T-cell lymphocytes were next identified by CD3 expression on the CD45 versus CD3 dual plot (labelled CD3+ cells; red). Colour compensation was required to correctly identify this population on the CD45 versus SS dual plot, where together with mature B-cells, they characteristically have the brightest CD45 expression and lowest complexity of all leucocytes. This involved using the “Pre-B_Plot Sheet” and slight manipulation of the CD3/CD81, CD3/CD34, CD3/CD10 and CD123/CD3 dual plots, with careful correlation on the “Analysis_haematogones panel” to ensure no erroneous background fluorescence was introduced. • NSB was then applied using the ungated CD123 versus CD19 dual plot (labelled NSB; black). This specific dual plot was included following a pattern of NSB noted on verification analysis of these fluorochromes for leucocyte FcRs. • Lastly, eosinophils were identified using the CD19 versus CD45 dual plot (labelled Eosinophils; teal) by their autofluorescence in violet light and distinct separation from the leucocyte populations highlighted in previous steps. Dual plots were then analysed using the stand-alone discriminatory markers (CD19, CD58, CD81 and CD123) against the primary (CD45 and CD34) and secondary (CD10 and CD38) backbone markers and allowed for the identification of maturing precursor B-cells within a background of mature B-cells. “Rare event” analysis was incorporated to visually assist in identifying very small B-cell populations by effectively magnifying the cells without affecting its actual percentage within the sample, as described earlier in this research report. 14 An example of the immunophenotypic analysis of a normal paediatric sample with abundant haematogones may be found in Figure 3.4.1.4 (see Appendix D). The histogram plot included in this figure depicts the total B-cell population within the sample (using the CD45 versus CD19 gating strategy) and the variable antigenic expression and intensities of the precursor B-cells present (using the Boolean gating logic described above). 3.4.2. Study-specific bench procedure At the CMJAH NHLS flow cytometry laboratory, trained technologist personnel performed daily bench internal quality control (IQC) and formally processed samples for immunophenotypic analysis according to the laboratory’s standard operating procedures (SOPs), which may be provided on request. Beckman Coulter Navios flow cytometers were used, maintained and quality-controlled as per the manufacturer’s instructions and laboratory SOPs. Specialist haematopathologists on the flow cytometry bench reviewed all clinical history and laboratory parameters for samples received, determining whether samples would be processed for immunophenotypic analysis or not. This was done according to the Bethesda International Consensus guidelines on flow cytometric immunophenotypic analysis of haematolymphoid neoplasms, optimised and modified for local use.(51) Upon reception of suspected new or relapsed paediatric acute leukaemias, haematopathologists on the bench performed their routine immunophenotypic investigation using the established ClearLLab 10CTM and cytoplasmic panels for lineage assignation and further characterisation. Follow-up paediatric B-cell ALL samples generally required ClearLLab 10CTM B-cell/M2 panels for MRD analysis. The in-house immunophenotypic panel was requested as a secondary investigation in newly-diagnosed or relapsed paediatric B-cell ALL and their follow-up samples. This did not interfere with concurrent diagnostic practice, as only remnant patient BMA samples were utilised. 15 3.5. Data collection and patient records All newly-diagnosed and relapsed paediatric B-cell ALL patients were followed for the duration of their treatment in the specified study period. Patient record of diagnostic and follow-up flow cytometry and cPCR findings were reviewed using the NHLS Webview and TrakCare laboratory information systems. This information was documented using a password-protected Excel spreadsheet with patient identifiers removed. An example of a single patient’s diagnostic and follow-up information captured may be found in Table 3.5.1 in Appendix E. The completed spreadsheet may be provided on request. Further patient characteristics (e.g. age, gender and defined abnormality presence) were not evaluated and were beyond the scope of this research report. 4. DATA ANALYSIS 4.1. Part 1 In a prior validation study conducted in the CMJAH NHLS flow cytometry laboratory, CD58-FITC proved efficacious in differentiating haematogones from LAIPs and polyclonal B-cells in paediatric B- cell ALL using the Beckman Coulter DURAClone ALB (abnormal B-progenitor cells) panel.(52,53) Furthermore, although not formally included in a United Kingdom National External Quality Assurance (EQA) Service programme report in May 2021, our flow cytometry laboratory confirmed CD58-FITC’s consistent and expected level of expression in the identification of the prepared sample’s leukaemic cell population.(54) Specific to this study, the manufacturer-specified performance of discriminatory markers CD58-FITC and CD81-APC-H7 required verification by their inclusion in the in-house immunophenotypic panel, in which precursor B-cells at varying stages of development were identified in normal paediatric BMA samples (i.e. BMAs with an expected haematogone population and no identifiable target population). 20 remnant paediatric samples on which the ClearLLab 10CTM B-cell diagnostic panel was originally applied, were included in the verification component of this study (see Table 4.1.1; Appendix F). 16 Instrument and method precision was established by running 10 separate, consecutive acquisition runs (of at least 50 000 events each) on residual paediatric BMA samples from two paediatric patients, one with no disease (i.e. rich in haematogones) and one with disease (i.e. rich in B-cell ALL leukaemic cells). The in-house haematogone data analysis protocol was applied, tailored and superimposed (see description below) on consecutive samples for which the analysed data may be reviewed on request. A carry-over assessment was conducted for the diseased sample to confirm minimal carry-over between samples run on the Beckman Coulter Navios flow cytometer machine as per international standards (see Figure 4.1.1; Appendix G). Individual lowest limit of quantification (LLOQ) assessment of the ClearLLab 10CTM B-cell/M2 and the in-house immunophenotypic panels was determined using a serial dilution approach with phosphate- buffered saline (PBS) in a paediatric BMA sample with a high B-cell leukaemic burden. Beads were added to each dilution to confirm the persistent number of white cell events in each sequential sample as the diluent (PBS) was added. Fixed quantities of the appropriate panels were then added to each labelled dilution, processed according to the flow cytometry laboratory SOPs and analysed by application of panel-specific data analysis protocols using the Beckman Coulter KaluzaCTM software system. Owing to limited sample quantity and time constraints, 20 000 events were analysed for each dilution. Individual panel precision was confirmed by conducting five consecutive analyses of the respective panel dilution in which no leukaemic cells were detectable. Table 4.1.2 in Appendix H may be reviewed for further information. 4.2. Part 2 The in-house immunophenotypic panel was added to each patient’s diagnostic remnant BMA sample, following which individual case analysis involved tailoring the diagnostic gates of the in-house haematogone data analysis protocol to more accurately identify the leukaemic cell population, including “rare event” analysis where required. In the case of the addition of drop-in CD58-FITC and CD123-ECD (phycoerythrin-Texas Red conjugate) reagents to the established ClearLLab 10CTM B-cell tube in the 17 initial pilot stage of the study, the sample was analysed as per ClearLLab 10CTM B-cell data analysis protocol, with Kappa-FITC and CD10-ECD replaced by the discriminatory markers of interest. Figure 4.2.1 shows the diagnostic gating strategy used to identify a patient’s LAIP (CD19 versus CD10 dual plot gated with the “Precursor B-cell” Boolean logic described previously), that is small and non- complex (FS versus SS dual plot) showing aberrant dim CD45 expression (CD45 versus SS dual plot) and an early precursor B-cell immunophenotype. An example of the full immunophenotypic analysis of a patient’s diagnostic remnant BMA sample is shown in Figure 4.2.2 (see Appendix I). Figure 4.2.1. Example of a paediatric B-cell ALL with strategies employed to identify the LAIP. ALL, acute lymphoblastic leukaemia; LAIP, leukaemia-associated immunophenotype; CD, cluster of differentiation; SS, side scatter; FS, forward scatter; INT, integral; DX, diagnostic gate; Haem, haematogone. Follow-up samples were analysed either with the drop-in of the reagents by means of logical inference using the in-house haematogone data analysis protocol, or by the unique superimposition analysis strategy introduced in the CMJAH NHLS flow cytometry unit. The latter involved initial analysis of patient- unique LAIPs [defined by mean fluorescence intensity (MFI)] on diagnostic samples using the in-house haematogone data analysis protocol. This was followed by the creation of a patient-specific, tailored data analysis protocol into which follow-up sample data could be selected and dragged into; effectively being superimposed on their diagnostic LAIP. This strategy is only able to be employed in current state of the art flow cytometers such as the Beckman Coulter Navios instrument used in this study. 18 Figure 4.2.3 below shows an example of distinct clustering within dual plots incorporating the in-house haematogone data analysis protocol discriminatory markers, indicative of residual disease. Distinct clustering was determined qualitatively by visual identification, with “rare event” analysis applied in cases with very small numbers of residual leukaemic cells. Absolute count determination was beyond the scope of this research report. An example of the full immunophenotypic analysis of a single patient’s follow-up samples, both with (distinct clustering in the majority of diagnostic gates) and without residual disease (no distinct clustering present) may be found in Figures 4.2.4 and 4.2.5 respectively (see Appendix J). Figure 4.2.3. Example of a paediatric B-cell ALL residual disease with distinct clustering noted in the dual plots incorporating the discriminatory markers of interest in the in-house haematogone data analysis protocol. ALL, acute lymphoblastic leukaemia; CD, cluster of differentiation; DX, diagnostic gate; Haem, haematogones; FITC, fluorescein isothiocyanate; APC-H7, allophycocyanin-cyanine dye; ECD, phycoerythrin-Texas Red conjugate. Performance and predictive powers of the ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels as compared to cPCR were then determined, retaining binary outcome data for all eligible patients (see Tables 4.2.1 and 4.2.2; Appendix K). Due to quasi separation of points, qualitative multivariate analysis using a Firth bias-correction logistic regression model was performed to establish the association between the dependent binary outcome variable (cPCR) and independent exposure variables (individual immunophenotypic panels). Taking into consideration that some patients had more than one follow-up 19 sample, the first observation per patient was utilised for the purpose of estimating the area under the curve (AUC) measure, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the individual immunophenotypic panels. 5. RESULTS 5.1. Part 1 The in-house immunophenotypic precursor B-cell marker expression matched the expected variable expression in this leucocyte population. No abnormal B-cell populations (i.e. leukaemic B-cells) were identified in all samples tested. The additional markers included in the in-house immunophenotypic panel thus met the requirements for intended use and manufacturer performance specifications documented in applicable Beckman Coulter product specifications.(55) Of note, despite varying degrees of remnant sample haemodilution, NSB and requirement for colour compensation, the in-house haematogone data analysis protocol was still able to effectively visualise small populations of B-cells at various stages of maturation using “rare event” analysis allowing for visual magnification. Instrument and method precision was confirmed, with minimal carry-over between samples run on the Beckman Coulter Navios flow cytometer machine as per international standards (i.e. less than 5% after injection of a blank sample following the running of a high concentration diseased sample), documented in Table 4.1.1 (see Appendix G). The ClearLLab 10CTM B-cell/M2 panels both had a LLOQ at a 1:1024 dilution (0.001%), whilst the in-house immunophenotypic panel had a LLOQ at 1: 512 (0.02%). Lastly, individual panel precision was confirmed by conducting five consecutive analyses at their respective identified LLOQ. 5.2. Part 2 5.2.1 Overview of eligible patients The study fell short of the calculated sample size (n = 45) with a total of 40 eligible patients for final statistical analysis (see Figure 5.2.1.1). The sample size of 40 was sufficient enough to detect 90% 20 sensitivity and 85% specificity with a ± 15% precision. Reasons for ineligibility have been summarised in the figure below. Four patient samples were analysed for demonstration interest. Figure 5.2.1.1. Research study patient and sample eligibility. n, number; cPCR conventional polymerase chain reaction; CMJAH, Charlotte Maxeke Johannesburg Academic Hospital. 5.2.2. Qualitative method analysis Individual immunophenotypic panel sensitivity, specificity, PPV and NPV measures are presented in Table 5.2.2.1 below. When compared to cPCR, the ClearLLab 10CTM B-cell/M2 panels demonstrated low sensitivity (44%) and high specificity (90%), with a PPV and NPV of 0.7 and 0.8 respectively. In contrast, the in-house immunophenotypic panel demonstrated both a high sensitivity (81%) and specificity (100%), with a very high PPV (1.0) and NPV (0.9). Receiver Operating Characteristic (ROC) curve analysis further revealed an AUC of 0.7292 and 0.9583 for the ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels respectively (see Figure 5.2.2.1). These findings provide evidence that the in-house immunophenotypic panel is overall superior in performance and predictive power as compared to the ClearLLab 10CTM B-cell/M2 panels in the determination of MRD in paediatric B-cell ALL whilst awaiting the results of cPCR. Inappropriate sample received (8) Diagnostic or relapsed disease in which no in-house immunophenotypic panel and/or cPCR added (9) 78 samples (n = 40 patients) Incomplete sample collection (52) • No follow-up sample received (8) • No in-house immunophenotypic panel on follow-up samples, having had the panel added at diagnosis (41) • Insufficient remnant BMA sample (0) • Diagnostic sample analysed at a different flow cytometry unit (2) • Patient follows up at a hospital that does not submit their specimens to CMJAH flow cytometry unit (1) Mixed immunophenotype/lineage designation not possible at diagnosis (2) 71 samples (n = 30 patients) 21 Table 5.2.2.1. Sensitivity and specificity measures for ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels. ClearLLab 𝟏𝟏𝟏𝟏𝟏𝟏𝐓𝐓𝐓𝐓 B-cell/M2 panels In-house immunophenotypic panel Statistic Estimate Standard error 95% Confidence Interval Estimate Standard error 95% Confidence Interval Sensitivity 0.4444 0.1656 0.1198 0.7691 0.8182 0.1163 0.5903 1.0000 Specificity 0.9048 0.0641 0.7792 1.0000 1.0000 0.0000 1.0000 1.0000 PPV 0.6667 0.1925 0.2895 1.0000 1.0000 0.0000 1.0000 1.0000 NPV 0.7917 0.0829 0.6292 0.9541 0.9231 0.0523 0.8207 1.0000 M2, myeloid-2; PPV, positive predictive value; NPV, negative predictive value. Figure 5.2.2.1. Receiver Operating Characteristic curve used to assess the association between the binary outcome (cPCR) and independent exposure variables (ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels). cPCR, conventional polymerase chain reaction; M2, myeloid-2; ROC, Receiver Operating Characteristic; AUC, area under the curve. 5.2.3. Other noteworthy findings and observations All samples analysed contained sufficient remnant sample upon which the in-house immunophenotypic panel could be applied. Despite variable degrees of haemodilution, immunophenotypic analysis was 22 possible even with a low leukaemic burden at diagnosis or follow-up, the latter visualised and amplified graphically using “rare event” analysis. • Six samples (6/78; 8%) showed prominent NSB for which accurate analysis using the in-house immunophenotypic panel could not confirm immunophenotypic remission status. One sample (1/78; 1%) was too degenerate for any meaningful interpretation, with a similarly degenerate sample analysed using the ClearLLab 10CTM B-cell/M2 panels. • Overall, 16 samples (16/78; 19%) did not have specific constituents of the in-house immunophenotypic panel added at diagnosis or follow-up. This was due to intermittent unavailability (e.g. CD19-PC7, CD10-APC700 and CD3-PC5.5) with a short period when the primary backbone marker (CD45-Kro) was unavailable towards the end of the study. CD81- APC-H7 and CD123-ECD was not added in one follow-up sample owing to human error. Despite this technicality, the in-house immunophenotypic panel was able to reliably identify precursor B-cells and the LAIP in these samples by application of knowledge of normal B-cell maturation patterns, positive and negative internal controls and aberrant marker expression on leukaemic cells. The ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels demonstrated 100% concordance in the identification of the LAIP on all diagnostic samples, as well as the identification of background leucocyte populations (characteristic expression profile and internal negative or positive controls). After having identified the leukaemic CD19/CD10 co-expressing population, discriminatory markers CD58-FITC and CD81-APC-H7 of the in-house immunophenotypic panel were of most value, especially when interpreted in conjunction with CD34-APC and CD38-PE, with “rare-event” analysis applied when applicable. CD123-ECD was less informative in terms of distinguishing regenerating B- cell precursors from residual disease. • Five samples (5/78; 6%) concurred with cPCR negativity for MRD, whilst the ClearLLab 10CTM B-cell/M2 panels suggested or could not exclude a small population of residual disease. The 23 main reason for the inability of exclusion of residual disease in the established panels were due to the absence of aberrant markers to better distinguish residual disease versus regenerating precursor B-cells. The in-house immunophenotypic panel is superior in this regard, capable of distinguishing the full maturation series based on appropriate discriminatory marker expression and level of brightness. • Eight samples (8/78; 10%) concurred with cPCR positivity for MRD, whilst ClearLLab 10CTM B-cell/M2 panels were not suggestive of residual disease or could not exclude residual disease due to the absence of aberrant markers described above. • Eight samples (8/78; 10%) showed out of range clonal products of uncertain clinical significance by cPCR, for which the concurrent in-house immunophenotypic panel confirmed no residual disease. • Two samples (2/78; 3%) showed cPCR negativity for MRD whilst both the ClearLLab 10CTM B-cell/M2 and in-house immunophenotypic panels showed immunophenotypic evidence of residual disease. • One sample (1/78; 1%) was extremely haemodilute, essentially resembling that of peripheral blood and both immunophenotypic panels were unable to exclude residual disease, whilst cPCR for MRD was positive for residual disease. • One sample (1/78; 1%) showed immunophenotypic evidence of two distinct leukaemic subpopulations, identified on both immunophenotypic panels. This specific case showed evidence of monocytic disease evolution, best detected by the CD45 versus SS dual plots of the in-house immunophenotypic panel. Further immunophenotypic information was not possible with this in-house haematogone data analysis protocol alone. • Four samples (4/78; 5%) expressed dim to negative leukaemic CD45 expression. By sound logic and inference, these populations were not excluded from analysis. The following was noted on the addition of the in-house immunophenotypic panel to the samples of demonstration interest: 24 • One sample showed a polyclonal IgH gene re-arrangement at diagnosis, and thus cPCR was not required on subsequent follow-up samples unless relapsed disease was suspected. The in-house immunophenotypic panel was added to four of this patient’s follow-up samples, one of which showed prominent NSB and could not reliably be interpreted. The remaining three samples, however confirmed no residual leukaemic cells, with variable precursor B-cells present in keeping with no MRD. • Two samples were in consensus with the CMJAH NHLS flow cytometry laboratory EQA proficiency testing scheme and accreditation programme in the identification of a diagnostic sample with B-cell ALL. The first sample incorporated the drop-in of CD58-FITC, whilst the second sample incorporated the full in-house immunophenotypic panel. These findings show comparability to international standards for flow cytometric analysis of diagnostic B-cell ALL. 5.3 Conclusive remark This work confirms the findings of previous studies, where discriminatory marker CD58-FITC in combination with backbone informative markers demonstrates both superior diagnostic and monitoring utility in paediatric B-cell ALL. Furthermore, the study suggests that the in-house immunophenotypic panel is both superior to the established ClearLLab 10CTM B-cell/M2 panels, and compares to cPCR for MRD determination, supporting its introduction as a LDT into routine flow cytometry laboratory practice in this patient population. 6. DISCUSSION Risk stratification and prognostication is central to paediatric B-cell ALL management, allowing for appropriate initial regimen choice, individualised treatment consideration and potential future HSCT planning.(2,6) As an important prognostic factor in this patient population, MRD status informs the clinician about remaining disease burden after induction therapy using institution-specific, modified BFM protocols, reflecting chemosensitivity and treatment response (15,17). MRD status also forms a major 25 predictor of overall and disease-free survival, with the potential for earlier clinical decision-making pertaining to regimen modification (i.e. therapy minimisation versus intensification, or HSCT eligibility evaluation), improving morbidity and mortality.(2,6,9,11,18,56) cPCR for the identification of B-cell IgH gene re-arrangements as well as flow cytometry play well- established roles in MRD determination in paediatric B-cell ALL, the former being the most studied and superior molecular technique in the absence of defined genetic aberrations.(25) In comparison to cPCR, flow cytometry has a quicker turnaround time, offering timeous diagnostic reporting to facilitate appropriate patient management. However, international consensus offers no one optimal MFC panel that exists in terms of minimising cost and spectral overlap, whilst maximising relevant immunophenotypic information obtained.(14,38) Panel selection may thus be individualised as a LDT, dependent on resource and expertise availability whilst ensuring retention of the predictive power of larger panels.(13,21,44,45) This study aimed to determine whether an in-house immunophenotypic panel containing the discriminatory CD58-FITC compared to cPCR in the detection of paediatric B-cell ALL MRD, with concurrent comparison to the validated ClearLLab 10CTM B-cell/M2 panels used at the CMJAH NHLS flow cytometry laboratory. The study found that the in-house immunophenotypic panel was overall superior in performance and predictive power as compared to its counterpart, the latter demonstrating a notable low sensitivity in correctly identifying residual disease when present, largely owing to the absence of aberrant markers on LAIPs at diagnosis. By application of its highly specific Boolean gating strategy in the identification of diagnostic patient-specific LAIPs, the in-house immunophenotypic panel offers an attractive, comparable alternative in MRD determination in this patient population whilst awaiting PCR findings. Furthermore, adoption of this LDT into routine flow cytometry laboratory practice raises the possibility of earlier clinical decision-making with potential improvement of morbidity and mortality outcomes. This is particularly an incentive within resource-constrained sub-Saharan Africa where feasibility of performing one modality over the other is a reality.(2,6,9,11,18) 26 Many additional observations of clinical importance were made during the construct of this research report. The samples of demonstration interest showed both comparability to international standards for flow cytometric analysis of diagnostic B-cell ALL, and clinical utility in cases with polyclonal IgH gene re-arrangements. This study may form the backbone for future MFC validation studies in our flow cytometry laboratory (e.g. T-cell ALL MRD determination), as well as its application to different sample types for evidence of disease infiltration (e.g. pleural fluid, lymph node fine-needle aspiration and cerebrospinal fluid). Furthermore, long-term follow-up and correlation with the current knowledge of risk of relapsed disease in this patient population may be considered using this data.(56) Lastly, in addition to current literature these findings may advocate for the development of a pre-titrated commercial lyophylised panel (much like the ClearLLab 10CTM panel), obviating the need for flow cytometric titration and technical effort in sample preparation.(35) In addition to the smaller sample size recruited, this study encountered limitations predominantly related to the pre-analytical stage of analysis. These included limited remnant patient sample for secondary analysis, delayed sample processing (e.g. if a sample was received over non-business working days) and technical error (e.g. single panel constituents erroneously not added or due to reagent unavailability). These difficulties are not unique to this study and may occur in general flow cytometry laboratory work practice.(35) However, despite these technicalities, the in-house immunophenotypic panel was still able to overall reliably detect the presence or absence of the LAIP or precursor B-cells. The following solutions were instituted in an attempt to overcome these limitations experienced: the application of “rare event” analysis; inclusion of “NSB” into the Boolean gating strategy; real-time error correction and monitoring of reagent stock, and alternative marker assessment to identify the LAIP or precursor B-cell populations. The future of B-cell ALL MRD analysis is promising, with various novel modalities in study including high throughput sequencing MRD technologies and artificial intelligence MFC machine learning. The former includes next generation flow cytometry and sequencing, PCR-based high throughput sequencing of IgH gene re-arrangements, and third generation digital droplet cPCR.(10,11,13,18,27,57) The latter has been 27 hypothesised to negate the need for calibration curve analysis, whilst still accurately quantifying DNA target sequences.(13,17,25,57) The breakthrough phenomenon of artificial intelligence machine learning in the context of haematopathology is increasingly being recognised, with advancing algorithms identifying common immunophenotypic profiles, whilst uncovering populations and subpopulations undetectable by subjective human analysis and streamlining workflow.(58–61) These future modalities address the need for fast, standardised technologies with greater MRD diagnostic sensitivity within this patient population, however still require large scale validation studies, their feasibility in resource-constrained LMICs remaining reserved.(4,10,11,12,13,22,63) 7. ACKNOWLEDGEMENTS I would like to thank my supervisors in the development of this research report, the collaborative findings of which are hoped to be of value for the flow cytometry and referral paediatric oncology departments of the CMJAH complex. Professor Glencross’ expertise in the field of flow cytometry was invaluable in my understanding of the molecular modality. Professor Geel’s expert scientific writing skills were of particular value in the construct of the research report. Furthermore, I am grateful for the academic input, experimentation and laboratory skills of Dr Denise Lawrie, CMJAH NHLS flow cytometry laboratory manager and her team of technologists. Finally, I wish to thank Dr Kennedy Otwombe, associate professor and statistician at the CHBAH Perinatal HIV Research Unit for his expert assistance in the statistical component of this research study. 8. RESEARCH FUNDING Funding for this research was covered by DK Glencross Research Incentive Funds and other Wits Health Consortium managed funds. 28 9. ETHICS AND RESEARCH APPROVAL DOCUMENTS All pilot work conducted prior to formal MMed candidate ethics clearance was covered under supervisor Professor Deborah Glencross’ blanket ethic clearance for the evaluation and validation of kits, reagents, data analysis protocols and methods (M1704129). Formal ethics approval for the current research study was obtained from the University of the Witwatersrand’s Human Research Ethics Committee (M220153). Permission to use NHLS facilities and the CMJAH flow cytometry unit’s Beckman Coulter KaluzaCTM software system was obtained from the NHLS Academic Affairs and Research Office (PR2222608) and head of department respectively. Registration with the National Health Research Database (GP_202203_014) and formal approval to conduct research from the CMJAH chief executive officer was further obtained. 10. REFERENCES 1. Arber DA, Orazi A, Hasserjian R, Thiele J, Borowitz MJ, Le Beau MM, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016 May 19;127(20):2391–405. 2. Terwilliger T, Abdul-Hay M. Acute lymphoblastic leukemia: a comprehensive review and 2017 update. Blood Cancer J. 2017 Jun;7(6):e577–e577. 3. Dong Y, Shi O, Zeng Q, Lu X, Wang W, Li Y, et al. Leukemia incidence trends at the global, regional, and national level between 1990 and 2017. Exp Hematol Oncol. 2020 Jun 19;9(1):14. 4. Togo, B., Traore, F., Doumbia AK., Togo, P., Diall, H., Maiga, B, et al. Childhood acute lymphoblastic leukemia in sub Saharan Africa: 4 years’ experience at the pediatric oncology unit Bamako, Mali. J Child Adolesc Health. 2018;2(2). 5. Herbst M. Fact Sheet on Childhood Acute Lymphocytic Leukaemia (ALL). 2015. 6. Hefazi M, Litzow MR. Recent advances in the biology and treatment of B-cell acute lymphoblastic leukemia. Blood Lymphat Cancer Targets Ther. 2018 Sep 25;8:47–61. 29 7. Hendricks CL, Buldeo S, Pillay D, Naidoo A, Thejpal R, Rapiti N, et al. Comparing morphology, flow cytometry and molecular genetics in the assessment of minimal residual disease in children with B-acute lymphoblastic leukaemia (B-ALL). South Afr J Oncol. 2019 Oct 23;3(0):8. 8. Ketseoglou I. IdentiClone TM IGH Gene Clonality Assay for IGH gene rearrangement. 2019. 9. Stock, W., Estrov, Z. Up To Date. 2020 [cited 2021 May 25]. Clinical use of measurable residual disease detection in acute lymphoblastic leukemia - UpToDate. Available from: https://www.uptodate.com/contents/clinical-use-of-measurable-residual-disease-detection-in-acute- lymphoblastic-leukemia 10. van Dongen JJM, van der Velden VHJ, Brüggemann M, Orfao A. Minimal residual disease diagnostics in acute lymphoblastic leukemia: need for sensitive, fast, and standardized technologies. Blood. 2015 Jun 25;125(26):3996–4009. 11. Ikoma MRV, Beltrame MP, Ferreira SIACP, Souto EX, Malvezzi M, Yamamoto M, et al. Proposal for the standardization of flow cytometry protocols to detect minimal residual disease in acute lymphoblastic leukemia. Rev Bras Hematol E Hemoter. 2015 Dec;37(6):406–13. 12. Fuda F, Chen W. Minimal/Measurable Residual Disease Detection in Acute Leukemias by Multiparameter Flow Cytometry. Curr Hematol Malig Rep. 2018 Dec;13(6):455–66. 13. Kruse A, Abdel-Azim N, Kim HN, Ruan Y, Phan V, Ogana H, et al. Minimal Residual Disease Detection in Acute Lymphoblastic Leukemia. Int J Mol Sci. 2020 Feb 5;21(3):1054. 14. Theunissen P, Mejstrikova E, Sedek L, van der Sluijs-Gelling AJ, Gaipa G, Bartels M, et al. Standardized flow cytometry for highly sensitive MRD measurements in B-cell acute lymphoblastic leukemia. Blood. 2017 Jan 19;129(3):347–57. 15. Logan, A. Emerging and Practical Concepts and Controversies in Leukaemias. Global Leukaemia Academy Meeting; 2021 May 15. 16. Tembhare PR, Pg PGS, Ghogale S, Chatterjee G, Patkar NV, Gupta A, et al. A High-Sensitivity 10-Color Flow Cytometric Minimal Residual Disease Assay in B-Lymphoblastic Leukemia/Lymphoma Can Easily Achieve the Sensitivity of 2-in-106 and Is Superior to Standard Minimal Residual Disease Assay: A Study of 622 Patients. Cytometry B Clin Cytom. 2020;98(1):57–67. 17. Abou Dalle I, Jabbour E, Short NJ. Evaluation and management of measurable residual disease in acute lymphoblastic leukemia. Ther Adv Hematol. 2020 Jan 1;11:2040620720910023. 30 18. Della Starza I, Chiaretti S, De Propris MS, Elia L, Cavalli M, De Novi LA, et al. Minimal Residual Disease in Acute Lymphoblastic Leukemia: Technical and Clinical Advances. Front Oncol [Internet]. 2019 Aug 7 [cited 2021 May 25];9. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6692455/ 19. Chernysheva O, Grivtsova LY, Popa A, Tupitsyn NN. B-Cell Precursors: Immunophenotypic Features in the Detection of Minimal Residual Disease in Acute Leukemia [Internet]. Normal and Malignant B-Cell. IntechOpen; 2019 [cited 2021 Jun 30]. Available from: https://www.intechopen.com/books/normal-and- malignant-b-cell/b-cell-precursors-immunophenotypic-features-in-the-detection-of-minimal-residual- disease-in-acute-le 20. Shaver, A. International Clinical Cytometry Society. [cited 2021 Feb 21]. Selecting a B- ALL MRD panel. Available from: https://cytometry.org/public/newsletters/eICCS-5-3/article3.php 21. Keeney M, Wood BL, Hedley B, DiGiuseppe JA, Stetler-Stevenson M, Paietta E, et al. Experience with MRD Testing in B- ALL By Flow Cytometry Does Not Prevent Interpretative Discordance. Blood. 2016 Dec 2;128(22):2907–2907. 22. Don MD, Lim W, Lo A, Cox B, Huang Q, Kitahara S, et al. Improved Recognition of Hematogones From Precursor B-Lymphoblastic Leukemia by a Single Tube Flow Cytometric Analysis. Am J Clin Pathol. 2020 May 5;153(6):790–8. 23. Shin S, Hwang IS, Kim J, Lee KA, Lee ST, Choi JR. Detection of Immunoglobulin Heavy Chain Gene Clonality by Next-Generation Sequencing for Minimal Residual Disease Monitoring in B-Lymphoblastic Leukemia. Ann Lab Med. 2017 Jul;37(4):331–5. 24. Schwab C, Harrison CJ. Advances in B-cell Precursor Acute Lymphoblastic Leukemia Genomics. HemaSphere. 2018 Aug;2(4):e53. 25. Correia RP, Bento LC, de Sousa FA, Barroso R de S, Campregher PV, Bacal NS. How I investigate minimal residual disease in acute lymphoblastic leukemia. Int J Lab Hematol. 2021;43(3):354–63. 26. Instructions for use: IdentiClone IGH Gene Clonality Assay [Internet]. 2022 [cited 2023 Apr 4]. Available from: https://invivoscribe.com/uploads/products/instructionsForUse/280255.pdf 27. Wood BL. Flow cytometry in the diagnosis and monitoring of acute leukemia in children. J Hematop. 2015 Sep;8(3):191–9. 28. Matutes E, Morilla, R., Morilla, AM. Immunophenotyping. In: Dacie and Lewis: Practical Haematology. Eleventh Edition. Elsevier; 2011. p. 353–71. 31 29. Hupp MM, Bashleben C, Cardinali JL, Dorfman DM, Karlon W, Keeney M, et al. Participation in the College of American Pathologists Laboratory Accreditation Program Decreases Variability in B- Lymphoblastic Leukemia and Plasma Cell Myeloma Flow Cytometric Minimal Residual Disease Testing: A Follow-up Survey. Arch Pathol Lab Med. 2021 Mar 1;145(3):336–42. 30. Khumalo, NM. Standard operating procedure HAE0005: Flow cytometry theory. 2020. 31. Dworzak MN, Buldini B, Gaipa G, Ratei R, Hrusak O, Luria D, et al. AIEOP-BFM Consensus Guidelines 2016 for Flow Cytometric Immunophenotyping of Pediatric Acute Lymphoblastic Leukemia. Cytometry B Clin Cytom. 2018;94(1):82–93. 32. Johansson U, Bloxham D, Couzens S, Jesson J, Morilla R, Erber W, et al. Guidelines on the use of multicolour flow cytometry in the diagnosis of haematological neoplasms. Br J Haematol. 2014;165(4):455–88. 33. Glencross, DK. Standard operating procedure HAE0251: Clinical interpretation of a case using Flow Cytometry and Immunophenotyping, and including the use of Kaluza C analysis software. 2020. 34. Hedley BD, Cheng G, Keeney M, Kern W, Padurean A, Luider J, et al. A multicenter study evaluation of the ClearLLab 10C panels. Cytometry B Clin Cytom. 2021;100(2):225–34. 35. Glencross DK, Swart L, Pretorius M, Lawrie D. Evaluation of fixed-panel, multicolour ClearLLab 10C at an academic flow cytometry laboratory in Johannesburg, South Africa. Afr J Lab Med. 2022 Jul 15;11(1):1458. 36. Swart L, Pretorius M, Lawrie D, Glencross DK. Commercial DURAClone panels for extending the repertoire of multicolour immunophenotypic panels in an academic flow cytometry laboratory in South Africa. Afr J Lab Med. 2022;11(1):1720. 37. Huarriz A. Next Generation FlowTM solution to detect Minimal Residual Disease (MRD) in BCP-ALL patients. Webinar presented at: Cytognos Webinar Series; 2021 Jun 29. 38. Best Practices for Multiparametric Flow Cytometry - ZA [Internet]. [cited 2021 Jun 6]. Available from: //www.thermofisher.com/za/en/home/references/newsletters-and-journals/bioprobes-journal-of-cell- biology-applications/bioprobes-79/best-practices-multiparametric-flow-cytometry.html 39. Lim J, Petersen M, Bunz M, Simon C, Schindler M. Flow cytometry based-FRET: basics, novel developments and future perspectives. Cell Mol Life Sci. 2022 Mar 30;79(4):217. 32 40. Karawajew L, Dworzak M, Ratei R, Rhein P, Gaipa G, Buldini B, et al. Minimal residual disease analysis by eight-color flow cytometry in relapsed childhood acute lymphoblastic leukemia. Haematologica. 2015 Jul;100(7):935–44. 41. Borowitz M, Chan J, Downing J, Le Beau M, Arber D. Precursor lymphoid neoplasms. In: WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Revised 4th Edition. Lyon: International Agency for Research on Cancer; 2017. p. 200–9. 42. Jaffe, ES., Campo, E., Harris, NL., Pileri,SA., Stein, H., Swerdlow, SH. Introduction and overview of the classification of the lymphoid neoplasms. In: WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Revised 4th Edition. Lyon: International Agency for Research on Cancer; 2017. p. 190–2. 43. Keeney M, Wood BL, Hedley BD, DiGiuseppe JA, Stetler-Stevenson M, Paietta E, et al. A QA Program for MRD Testing Demonstrates That Systematic Education Can Reduce Discordance Among Experienced Interpreters. Cytometry B Clin Cytom. 2018 Mar;94(2):239–49. 44. Shaver A, Greig B, Mosse C, Seegmiller A. B-ALL Minimal Residual Disease Flow Cytometry: An Application of a Novel Method for Optimization of a Single-Tube Model. Am J Clin Pathol. 2015 May 1;143:716–24. 45. Rozanne Adams. Introduction into Multiparameter Panel Design. Virtual Seminar presented at: Flow Cytometry Seminar BD Life Sciences; 2021 Jun 9. 46. Das N, Gupta R, Gupta SK, Bakhshi S, Malhotra A, Rai S, et al. A Real-world Perspective of CD123 Expression in Acute Leukemia as Promising Biomarker to Predict Treatment Outcome in B-ALL and AML. Clin Lymphoma Myeloma Leuk. 2020 Oct 1;20(10):e673–84. 47. Lee RV, Braylan RC, Rimsza LM. CD58 expression decreases as nonmalignant B cells mature in bone marrow and is frequently overexpressed in adult and pediatric precursor B-cell acute lymphoblastic leukemia. Am J Clin Pathol. 2005 Jan;123(1):119–24. 48. Vaughan J, Bouwer N, Willem P, Wiggill T, Hodkinson K. The translocation t(1;19)(q23;p13) (TCF3/PBX1 fusion) is the most common recurrent genetic abnormality detected amongst patients with B- cell lymphoblastic leukaemia in Johannesburg, South Africa. South Afr J Oncol. 2021 Jul 6;5(0):6. 49. Pretorius, M., Glencross, DK. Flow Cytometry CMJAH Registrar Training Notes for ClearLLab/ Duraclone/ and CMJAH In-House Panel Analysis. 2021. 33 50. Andersen MN, Al-Karradi SNH, Kragstrup TW, Hokland M. Elimination of erroneous results in flow cytometry caused by antibody binding to Fc receptors on human monocytes and macrophages. Cytometry A. 2016;89(11):1001–9. 51. Davis BH, Holden JT, Bene MC, Borowitz MJ, Braylan RC, Cornfield D, et al. 2006 Bethesda International Consensus recommendations on the flow cytometric immunophenotypic analysis of hematolymphoid neoplasia: medical indications. Cytometry B Clin Cytom. 2007;72 Suppl 1:S5-13. 52. Glencross D. Validation report for the Beckamn Coulter Duraclone RE: ALB tube. Charlotte Maxeke Johannesburg Academic Hospital: Universoty of the Witwatersrand; 2020 Aug. (Extended workup for investigation of immature B- cells). 53. Glencross D, Pretorius M, Swart L, Lawrie D. Commercial multicolour, fixed, lyophilised DURAClone RE ALB, CLB & PC panels can extend the repertoire of fixed multi-colour, diagnostic marker panels. AJLM Peer Rev. 2020; 54. Leukaemia Immunophenotyping (Part 1) Programme. United Kingdom: Sheffield Teaching Hospitals NHS Foundation Trust; 2021 Jul p. 9. (Leucocyte Immunophenotyping). Report No.: 212201. 55. Product Listings and Catalogs [Internet]. [cited 2023 Jul 16]. Available from: https://www.beckman.it/reagents/coulter-flow-cytometry/antibodies-and-kits/single-color- antibodies/product-listing 56. Campbell M, Kiss C, Zimmermann M, Riccheri C, Kowalczyk J, Felice MS, et al. Childhood Acute Lymphoblastic Leukemia: Results of the Randomized Acute Lymphoblastic Leukemia Intercontinental- Berlin-Frankfurt-Münster 2009 Trial. J Clin Oncol [Internet]. 2023 May 4 [cited 2023 Nov 8]; Available from: https://ascopubs.org/doi/pdf/10.1200/JCO.22.01760?role=tab 57. Starza ID, Nunes V, Cavalli M, Novi LAD, Ilari C, Apicella V, et al. Comparative analysis between RQ- PCR and digital-droplet-PCR of immunoglobulin/T-cell receptor gene rearrangements to monitor minimal residual disease in acute lymphoblastic leukaemia. Br J Haematol. 2016;174(4):541–9. 58. Krause SW. On Its Way to Primetime: Artificial Intelligence in Flow Cytometry Diagnostics. Cytometry A. 2020;97(10):990–3. 59. Lin E, Fuda F, Luu HS, Cox AM, Fang F, Feng J, et al. Digital pathology and artificial intelligence as the next chapter in diagnostic hematopathology. Semin Diagn Pathol. 2023 Mar;40(2):88–94. 34 60. Fuda F, Chen M, Chen W, Cox A. Artificial intelligence in clinical multiparameter flow cytometry and mass cytometry–key tools and progress. Semin Diagn Pathol. 2023 Mar 1;40(2):120–8. 61. Hennig H, Rees P, Blasi T, Kamentsky L, Hung J, Dao D, et al. An open-source solution for advanced imaging flow cytometry data analysis using machine learning. Methods San Diego Calif. 2017 Jan 1;112:201–10. 62. Patkar N, Alex AA, B B, Ahmed R, Abraham A, George B, et al. Standardizing minimal residual disease by flow cytometry for precursor B lineage acute lymphoblastic leukemia in a developing country. Cytometry B Clin Cytom. 2012;82B(4):252–8. APPENDIX A Table 1.4.1. Normal maturation of the B-cell. Major immunophenotypic markers are included for the purposes of this research report. * c, cytoplasmic; CD, cluster of differentiation; nTdT, nuclear terminal deoxynucleotidyl transferase; HLA-DR, human leucocyte antigen-DR isotype; PAX5, paired-box containing protein 5; BSAP, B-cell lineage-specific activator protein; μ, mu; Ig, immunoglobulin. ** Orange: denotes generalised antigen expression *** Grey: denotes non-antigen expression **** White: denotes antigenic expression at specific stages of maturation Bone marrow Lymph node Pro-B-cell Pre-B-cell Immature B- cell Mature Naive B- cell Germinal Centre B-cell Marginal Zone B-cell Plasma cell CD34 +/- CD34 HLA-DR CD45 nTdT +/- nTdT CD10 CD10 CD19 Dimmer CD19 +/- CD20 Increasing CD20 and cCD22 CD23 Bright CD38 Increasing CD38 Weak CD58 +/- CD79a CD79a CD81 +/- CD123 +/- PAX5/BSAP PAX5/BSAP CD138 c𝜇𝜇 chain Ig Surface Kappa/Lambda light chain (dependent on antigen exposure) 35 APPENDIX B Table 1.4.2. Useful markers to differentiate between LAIP and haematogone populations within B-cell ALL MRD bone marrow samples. Relevant marker considerations LAIP (and subclones if present) Maturing haematogone population CD34 o Marker of immaturity o Not lineage specific  Persistent expression suggests asynchronous maturation  May lack CD34 expression  May or may not be expressed  Lost relatively early in B-cell maturation CD45 o Marker for all leucocytes o Common initial gating strategy  Presents as a discrete population within the negative-to-dim CD45 gate  May diminish with clonal evolution or disease relapse  Presents as a non-discrete population within the negative-to-dim CD45 gate, indicating a maturing population  Dimmer than background normal polyclonal B-cells  Brighter than LAIP expression nTdT o Associated with early B- and T-cell development o Not specific for ALL  Useful only to assist in immaturity determination at initial diagnosis  May or may not be expressed CD10 o Expression is lost with B-cell maturation as cells acquire surface Ig expression o Re-expressed at the germinal centre stage of B-cell development  Brighter than normal B-cell precursors  Persistent expression suggests asynchronous maturation  Certain subtypes may have absent expression and yield early prognostic value  Decreased expression CD20 o Defines an early pre-B-cell o PAX5 is considered more sensitive and specific to detect pre-B-cells immunohistochemically  May or may not be expressed  Earlier expression may suggest asynchronous maturation  Continued and brighter expression with maturation CD22 o Expression is restricted to B- cells  Brighter than normal B-cell precursors  Earlier expression may suggest asynchronous maturation  Decreased expression CD38 o Widely expressed in most leucocytes with its intensity largely correlating with the degree of cellular activation  Variable expression and decreased as compared to haematogones  Uniform bright expression CD58 o CD38/CD58 co-expression is most frequently used to identify LAIP (insufficient power if used in isolation)  Bright and overexpressed  Correlates with CD10 identification of LAIP in terms of brightness and asynchronous maturation  Expression is lost in early maturation CD81 o Widely expressed on immune cells  Dim and underexpressed  Uniform bright expression CD123 o Low or negative expression in primitive haemopoietic cells  Bright and overexpressed  Discordant expression favours haematogones at varying stages of 36 MRD, measurable residual disease; CD, cluster of differentiation; ALL, acute lymphoblastic leukaemia; nTdT, nuclear terminal deoxynucleotidyl transferase; LAIP, leukaemia-associated immunophenotypes; PAX5, paired- box containing protein 5; cPCR, conventional polymerase chain reaction; Ig, immunoglobulin.  Concordant expression favours LAIP (CD34+/CD123+; CD34-/CD123-) maturation (CD34+/CD123-) and (CD34-/CD123+) Surface Igs o Used in cPCR for clonal determination  Very rarely express surface kappa and lambda light chains  Absent surface expression 37 APPENDIX C 38 39 Figure 3.4.1.1. In-house haematogone data analysis protocol using the Beckman Coulter KaluzaCTM software system. CD, cluster of differentiation; SS, side scatter; FS, forward scatter; NSB, nonspecific binding; Mo, monocytes; Gr, granulocytes; Ly, lymphocytes; Haem, haematogones; DX, diagnostic gate; FITC, fluorescein isothiocyanate; APC-H7, allophycocyanin-cyanine dye. ∗ Abbreviations of the non-assessed markers are not further described ∗ The blue diagnostic gates are discussed in the research report under "Data analysis" - Part 2 (pages 16 - 18) KEY 40 A