BCR-ABL Assay Sensitivity of MR4.5 Achieved in >90%, and MR5 in >75% of Samples, through mRNA Selection before qRT-PCR

Background

An increasing number of CML patients (pts) participate in trials of drug cessation. Deep molecular response (DMR) such as MR4.0 or MR4.5 is thought to be a necessary pre-condition to treatment free remission (TFR), and act as inclusion criteria for most TFR studies. Performing qRT-PCR assays that can consistently detect BCR-ABL at ≥4.5 log sensitivity (sens) below the standardised baseline is challenging for many labs. Undetectable (undet) BCR-ABL from assays with inadequate sensitivity may lead to inappropriate or premature cessation attempts. The associated failure may jeopardise future TFR success, and is emotionally traumatic for some pts. Furthermore, qRT-PCR with increased sens may allow for examination of potential correlations between very deep molecular responses (eg MR5 - MR6) and TFR success. Standard qRT-PCR methods for BCR-ABL detection conventionally use total RNA. We aimed to enhance sensitivity within the framework of the standardised qRT-PCR assay through mRNA selection and maximising input.

Method

Total RNA was extracted using Trizol from peripheral blood samples (159 samples from 116 CML pts mostly with DMR, and 58 known p210 BCR-ABL negative samples). BCR-ABL transcripts were quantified by qRT-PCR. The optimised standard method (StdM) was as described, using 2-10ug of total RNA  as starting material for reverse transcription (RT) (Branford BJH 1999:107(3);597-99). In the experimental method (ExpM), mRNA was concentrated from 5.1-59.5ug (median 18.8ug) of archival RNA from the corresponding sample with oligo-dT-bound magnetic beads (Dynabeads®), prior to RT. The same QPCR protocol was then used for both methods. BCR-ABL e13a2 and e14a2 were measured in separate reactions; BCR was quantified as the control gene, from which assay sens was calculated.

Results

ExpM was performed on 58 samples from non-CML pts; 98% had <15 BCR-ABL transcripts (median 0 and 1 copies of e14a2 and e13a2 respectively). Fifteen transcripts was established as the lower limit of detection. Of the 159 samples from CML pts, 120 had undet BCR-ABL (75%) with StdM (median sens 4.6; 3.7-5.2). Of the 120 undet samples, 33 (28%) had detectable BCR-ABL with ExpM at levels from (0.0002-0.008%).  Detection of low level residual disease was enabled by increased assay sens of 0.4-1.5 log. StdM detected BCR-ABL in 39/159 samples, 35 also had detectable BCR-ABL with ExpM. Numerically, qRT-PCR results from ExpM showed an excellent correlation with StdM (R2 = 0.93, p<0.001; Fig A). StdM detected disease in 4 samples negative for BCR-ABL by ExpM – 3/4 of the archival RNA had undet BCR-ABL on retesting with StdM; 1 sample had been exhausted.

Analysing all 217 samples, assay sens increased from a median of 4.5 log (3.7-5.3) with StdM to 5.3 log (3.0-6.4) with ExpM (Fig B), suggesting a maximum theoretical detection limit of 0.00005%. This increase of 0.8 log is equivalent to a ~6 fold increase in sens. With ExpM, 91% of assays had sens ≥4.5 log and 78% ≥5 log; vs StdM: 61% ≥4.5 log and 9% ≥ 5 log (Fig C).

Samples were available for 21 pts with >12 months follow up after TKI cessation: all 21 had undet BCR-ABL by StdM at cessation, whereas 4/21 had detectable BCR-ABL by ExpM. Three of these 4 were among 13/21 pts with molecular recurrence (loss of MMR; Fig D shows a time course for 1 pt). 

Conclusion

We have demonstrated significant improvement of the qRT-PCR method by incorporation of a relatively simple procedure. Selectively concentrating mRNA before qRT-PCR increased assay sens by a median of 0.8 log, equivalent to ~6 fold increase, with a theoretical detection limit of 0.00005%. Our method needs no specialised equipment (eg a digital PCR platform) or additional expertise and only marginally increases labour and cost. This method may therefore be feasibly integrated into routine practice, and will be particularly valuable in labs where routinely achieving sens at ≥MR4 is problematic. Avoiding the need for a pre-amplification step used in other highly sensitive qRT-PCR protocols circumvents potential contamination and improves accuracy of quantification. The protocol allowed >90% of samples to achieve ≥MR4.5 sensitivity, a clinically relevant level for qualification in many TKI cessation studies. The prognostic factors for successful TKI cessation remain elusive. The ability to detect BCR-ABL at the increased sens demonstrated in our method requires further evaluation to determine its prognostic value for in the TFR setting.