Evaluating reverse transcriptase efficiency in CML: Consequences for deep molecular response classification and treatment-free remission decisions

Abstract

Abstract Background In chronic myeloid leukemia (CML), deep molecular response (DMR) is defined as molecular response 4 (MR4) or deeper. If BCR::ABL1 is undetectable the molecular response is determined by the number of control gene (CG) copies—typically ABL1 or GUSB—detected by RT-qPCR. RT-qPCR remains the gold standard for monitoring measurable residual disease (MRD). Sensitivity thresholds are set as follows: MR4 requires 10,000–31,999 ABL1 copies (or 24,000–76,999 GUSB copies, reflecting its 2.4-fold higher expression); MR4.5 requires 32,000–99,999 ABL1 copies or 77,000-239,999 GUSB copies; and MR5 requires ≥100,000 ABL1 or ≥240,000 GUSB copies. The efficiency of reverse transcriptase (RT) enzymes used during cDNA synthesis can significantly impact transcript quantification and thereby MR classification. Aim As part of the European Treatment and Outcome Study for CML 2024 (EUTOS 2024) initiative, this study aimed to evaluate the influence of various RT enzymes on assay sensitivity and MR classification—an important consideration for treatment discontinuation decisions, as outlined in the European LeukemiaNet (ELN) 2025 guidelines. Methods Samples from 12 CML patients in deep molecular response (DMR) — 7 with major and 5 with atypical BCR::ABL1 transcripts — were analyzed. Equal amounts of RNA, within the range recommended by the respective manufacturers, were reverse-transcribed using seven different reverse transcriptase (RT) enzymes: SuperScript II (SSII), GoScript, MMLV, High Capacity, VILO, Clara, and GoTaq. Quantification of ABL1, GUSB, and BCR::ABL1 transcripts was performed using standardized RT-qPCR protocols. Therefore, the only variable in this analysis was the choice of RT enzyme. Results Using SSII, the expected GUSB/ABL1 expression ratio was confirmed (mean 2.57, SD 0.35; range 2.12–3.16), yielding consistent MR classification: 1 sample at MR4.5, 11 at MR5 for both CGs. High Capacity produced comparable results in 10 samples, with 2 classified at GUSB-MR5 but ABL1-MR4.5 (mean ratio 3.17, SD 0.53). MMLV showed slightly higher GUSB expression (mean ratio 3.42, SD 1.15) resulting in the classification of 5 samples as GUSB-MR5 but ABL1-MR4.5, while 6 remained at MR5 for both CGs and 1 sample at MR4.5 for both CGs. VILO and GoTaq exhibited markedly higher GUSB/ABL1 ratios—10.64 (SD 2.71) and 7.08 (SD 1.49), respectively—causing inconsistent MR classification across most samples. GoScript produced similar ABL1 and GUSB levels (mean ratio 1.3, SD 0.18) but overall lower transcript yields, leading to lower sensitivity. Clara showed highly variable performance and inconsistent GUSB/ABL1 ratios (SD 3.19, range 1.31-10.5), with 6/11 samples yielding false-negative BCR::ABL1 results. These findings indicate that ABL1 copy numbers are often insufficient, reflecting a lower analytical sensitivity. GUSB also showed suboptimal performance in most cases, with the exception of VILO, which tended to overestimate GUSB expression. Conclusions This study highlights the critical impact of reverse transcriptase choice on MR classification in CML, particularly in cases of undetectable BCR::ABL1. Only SuperScript II, High Capacity, and MMLV produced consistent and ELN-compliant results, including assessment of DMR. In contrast, GoTaq and GoScript generally produced insufficient copy numbers of both ABL1 and GUSB, resulting in reduced analytical sensitivity. VILO also yielded low ABL1 copy numbers but tended to slightly overestimate GUSB expression. Clara, due to inconsistent control gene transcription and a high risk of false-negative MRD results, appears unsuitable for clinical use in this context. We recommend that testing laboratories should validate their cDNA synthesis protocols to ensure that measured CG copy numbers fall within ELN-defined MR ranges, especially when evaluating patients for treatment-free remission.Support: EUTOS 2024, MH CZ – DRO (IHBT, 00023736)