Overall, the comparison of MRD results obtained by RQ-PCR and ddPCR showed a concordance rate of 79% [92/116 (P < 0.0001)] for samples classified as positive quantifiable (Q) or negative (NEG) by both techniques, whereas discordant results were detected in 21% (24/116) of samples. Sixteen were RQ-PCR-PNQ, 5 of which were ddPCR-Q and 11 of which were ddPCR-NEG; in the remaining 8 discordant FU samples, 5 were RQ-PCR-NEG/ddPCR-Q and 3 were RQ-PCR-NEG/ddPCR-PNQ.
Overall, the use of ddPCR significantly reduced the proportion of samples defined as PNQ—3 of 116 (2.6%) versus 16 of 116 (14%) (P = 0.003)—because quantified MRD in 5 of 16 RQ-PCR-PNQ samples and proved NEG in 11 of 16. Moreover, ddPCR also allowed to quantify the disease in 5 of 75 (6.6%) samples that were RQ-PCR-NEG, modifying MRD quantification by 14% (16/116) in samples defined RQ-PCR-PNQ or NEG (Figure 1 and Table 1).Figure 1Overall comparison of real-time quantitative PCR (RQ-PCR) and droplet digital PCR (ddPCR) minimal residual disease (MRD) results. Analyses were performed on 116 samples from 44 patients at four time points (TPs; TP1, TP2, TP3, and TP4). The correlation between MRD quantification by both methods shows a significant concordance (P < 0.0001). Sixty-seven samples were undetectable, and 25 samples were scored positive and quantifiable by both methods. In the gray shading, positive not quantifiable (PNQ) samples are highlighted. NEG, negative; Q, positive quantifiable.
Table 1Overall Comparison of RQ-PCR and ddPCR Plus NGS MRD Results
Analyses were performed on 116 samples at four time points (1, 2, 3, and 4) by RQ-PCR and ddPCR. Of 116 samples, 47 were also analyzed by NGS. NGS data are reported in italics.
ddPCR, droplet digital PCR; MRD, minimal residual disease; NEG, negative; NGS, next-generation sequencing; PNQ, positive not quantifiable; Q, positive quantifiable; RQ-PCR, real-time quantitative PCR.
In 47 of the 116 samples analyzed by RQ-PCR and ddPCR, MRD was also analyzed by NGS: 21 of 47 samples resulted Q, and 26 of 47 samples resulted NEG. The concordance rate between NGS and RQ-PCR and ddPCR was 75% (35/47 samples) and 92% (43/47 samples), respectively. Thus, NGS quantified the disease in 11 samples classified as PNQ or NEG by RQ-PCR, confirming the ddPCR results in 8 of 11 samples; and in 3 samples that were ddPCR-NEG, NGS was able to quantify the disease with an MRD level of 10−5. Moreover, in one RQ-PCR-PNQ sample, NGS was NEG; and in one sample, ddPCR-PNQ but RQ-PCR-NEG, NGS was NEG (Table 2).Table 2Comparison of RQ-PCR versus NGS and ddPCR versus NGS MRD Results
Analyses were performed on 47 samples at four time points (1, 2, 3, and 4). Concordant values between RQ-PCR and NGS and ddPCR and NGS are indicated in bold.
ddPCR, droplet digital PCR; MRD, minimal residual disease; NEG, negative; NGS, next-generation sequencing; PNQ, positive not quantifiable; Q, positive quantifiable; RQ-PCR, real-time quantitative PCR.
RQ-PCR, ddPCR, and NGS MRD Quantification at TP1, TP2, TP3, and TP4The RQ-PCR and ddPCR MRD analysis was performed at TP1, TP2, TP3, and TP4 for a total of 116 BM samples analyzed. At weeks 4 and 10 (TP1 and TP2, considered together to increase the sample size), the RQ-PCR and ddPCR analyses were performed on 78 BM samples. By RQ-PCR, 22 of 78 were Q, 9 of 78 were PNQ, and 47 of 78 were NEG, whereas by ddPCR, 29 of 78 were Q, 1 of 78 was PNQ, and 48 of 78 were NEG. Overall, at these TPs, the methods provided discordant results in 14 of 78 samples (18%): 6 were RQ-PCR-PNQ/ddPCR-NEG, 3 were RQ-PCR-PNQ/ddPCR-Q, 4 were RQ-PCR-NEG/ddPCR-Q, and 1 was RQ-PCR-NEG/ddPCR-PNQ. A subgroup of 31 of the 78 BM samples was analyzed also by NGS: 16 samples proved Q, and 15 were NEG. In the NGS group, 8 of 31 samples were RQ-PCR/ddPCR discordant (25.8%): 4 samples were RQ-PCR-NEG/ddPCR-Q and NGS resulted Q, 2 were RQ-PCR-PNQ/ddPCR-Q and NGS was Q, and 2 were RQ-PCR-PNQ/ddPCR-NEG while NGS was NEG and Q, showing that ddPCR and NGS provided reproducible results at the early TP.
At weeks 16 and 22 (TP3 and TP4), 38 BM samples were analyzed by RQ-PCR and ddPCR. By RQ-PCR, 3 resulted Q, 7 resulted PNQ, and 28 resulted NEG, whereas by ddPCR, 7 proved Q, 1 proved PNQ, and 30 proved NEG. The methods were discordant in 10 of 38 samples (26.3%): 2 RQ-PCR-PNQ/ddPCR-Q, 5 RQ-PCR-PNQ/ddPCR-NEG, 2 RQ-PCR-NEG/ddPCR-Q, and 1 RQ-PCR-NEG/ddPCR-PNQ. Of the 38 BM samples, 16 were also analyzed by NGS—11 NEG and 5 samples proved Q—with 3 being RQ-PCR/ddPCR discordant: 2 samples that were RQ-PCR-PNQ/ddPCR-Q resulted Q by NGS, and 1 that was RQ-PCR-NEG/ddPCR-PNQ was NEG by NGS.
Event-Free Survival of the Cohort Based on RQ-PCR/ddPCR/NGS Disease QuantificationOverall, 5 of 44 patients relapsed (11%), after a median follow-up of 33 months (range, 6 to 74 months). Three relapses occurred in patients who were PNQ/NEG by RQ-PCR but proved Q by ddPCR and/or NGS (60%), and in 2 patients defined Q by all methods (40%).
We evaluated event-free survival on the basis of the patients' disease levels at a single TP (TP1, TP2, and TP4). At TP1, by means of RQ-PCR, there were no statistically significant differences between Q (n = 15), NEG (n = 21), and PNQ (n = 5). At variance, evaluation by ddPCR resulted in the abrogation of PNQ cases, leading to an increase in both Q (n = 18) and NEG (n = 20) cases. As such, event-free survival was statistically different between the two subsets (P = 0.016) (Figure 2).Figure 2Event-free survival (EFS) by single time points (TPs). Kaplan-Meier plots depicting the EFS of patients stratified according to the presence of quantifiable, positive-nonquantifiable, or nondetectable disease by real-time quantitative PCR (RQ-PCR) and droplet digital PCR (ddPCR) at three TPs (TP1, TP2, and TP4). Pairwise P value for positive quantifiable (Q) arm is provided next to each related curve. PNQ, positive not quantifiable.
At TP2 and TP4, representing decisional time points for patients' allocation to allogeneic transplant, patients with Q levels by both RQ-PCR and ddPCR showed a significantly inferior outcome compared with those with PNQ or no detectable disease. In line with the results observed at TP1, ddPCR successfully quantified more patients who presented a relapse (four of five at TP2 and three of five at TP4) compared with RQ-PCR (two of five at TP2 and three of five at TP4). Finally, we classified patients based on the possibility to successfully quantify disease levels by their specific marker(s) longitudinally in more than one TP (marker tracing) (Figure 3). Multiple FUs allow us to measure the dynamics of patients' specific markers over time and to identify those with quantifiable disease in more than one TP. Of the 37 of 41 patients with more than one FU available, 5 were traceable by RQ-PCR (13%) and 10 were traceable by ddPCR (27%). Patients traceable by ddPCR had a significantly inferior event-free survival than those traceable by RQ-PCR (Figure 3). Marker tracing along greater than one TP was possible in four of five relapsed patients by ddPCR versus one of five by RQ-PCR (Figure 3).Figure 3Event-free survival (EFS) by marker tracing. Analysis was performed on 41 patients with more than one follow-up available. Left panel: Taking advantage of multiple longitudinal sampling, patients with quantifiable (green rectangles), positive not quantifiable (violet rectangles), and nondetectable disease (gray rectangles) have been identified at two or more time points (TPs; marker traced) by real-time quantitative PCR (RQ-PCR) and droplet digital PCR (ddPCR). Right panels: Kaplan-Meier plots depicting EFS of patients stratified by marker tracing. Global P value is provided on the bottom left of each plot. NEG, negative; PNQ, positive not quantifiable; Q, positive quantifiable.
DiscussionDuring RQ-PCR MRD monitoring in ALL, a consistent fraction of samples with low MRD levels cannot be properly quantified despite the use of the specifically developed EuroMRD guidelines.3van der Velden V.H. Cazzaniga G. Schrauder A. Hancock J. Bader P. Panzer-Grumayer E.R. Flohr T. Sutton R. Cave H. Madsen H.O. Cayuela J.M. Trka J. Eckert C. Foroni L. Zur Stadt U. Beldjord K. Raff T. van der Schoot C.E. van Dongen J.J.M. European Study Group on MRD detection in ALL (ESG-MRD-ALL)These data underline the greater accuracy and sensitivity of ddPCR that enabled us to identify an MRD signal also in low positive samples and, in particular, that ddPCR can provide a more robust and precise stratification for cases with a positivity <10−4, allowing us to distinguish true-positive cases from those defined as negative or PNQ by RQ-PCR.
Moreover, the value of 10−4 represents the most challenging cutoff both clinically and methodologically. At weeks 4 and 10 (TP1 and TP2), RQ-PCR and ddPCR were discordant in 14 samples, with 7 becoming Q by ddPCR. Similarly, at weeks 16 and 22 (TP3 and TP4), RQ-PCR and ddPCR were discordant in 10 samples, and in 4 ddPCR evaluation changed the MRD status compared with RQ-PCR (Q versus PNQ/NEG).
By survival analysis, both ddPCR and RQ-PCR were able to predict patients' relapse at TP2 and TP4 individually, with an advantage of ddPCR in capturing more events than RQ-PCR. Despite the small number of cases, this difference has important implications because the detection of Q levels of disease could lead to further therapeutic interventions in clinical trials and potentially rescue patients from relapse. On the other hand, it is likewise relevant to underline that no relapses were recorded in patients whose FU samples were defined as RQ-PCR-PNQ but proved NEG by ddPCR. These findings complement in part those previously reported by Raff et al,22Raff T. Gökbuget N. Lüschen S. Reutzel R. Ritgen M. Irmer S. Böttcher S. Horst H.A. Kneba M. Hoelzer D. Brüggemann M. Study Group G.M.A.L.L. Molecular relapse in adult standard-risk ALL patients detected by prospective MRD monitoring during and after maintenance treatment: data from the GMALL 06/99 and 07/03 trials. where relapses were observed also in cases with MRD within the quantitative range: as a matter of fact, in the present cohort, we documented three relapses in RQ-PCR-PNQ/NEG patients; noteworthy, these three cases were Q by ddPCR, thus showing the reliability of ddPCR over Q-PCR.Finally, when observing the disease level trend along multiple TPs (marker tracing), ddPCR was able to predict the occurrence of relapse. We believe this composite outcome measure is more precise than individual TPs (ie, TP2 or TP4) as it keeps into account the dynamic fluctuations of ALL over time.
Taken together, all these findings highlight, with the limit of the small sample size, the clinical advantage of ddPCR over Q-PCR in patients' stratification.
In a smaller number of samples (n = 47), we also analyzed MRD by NGS, which showed a concordance rate of 75% with RQ-PCR and of 92% with ddPCR. It confirmed the ddPCR quantification in samples classified as RQ-PCR-PNQ or NEG and allowed us to increase the rate of quantification, defining as positive and quantifiable three samples that resulted ddPCR-NEG (RQ-PCR-Q 21% versus ddPCR-Q 38% versus NGS-Q 45%). We are working on increasing the number of cases to analyze also by NGS. However, although NGS might in principle be more sensitive in quantifying MRD levels, it must be recalled that it requires higher amounts of DNA, not always available at follow-up TPs, and requires bioinformatics skill.
In an earlier publication, we performed a preliminary analysis by all three methods in 23 samples from 11 patients enrolled in two GIMEMA trials (6 cases from LAL1308 and 5 from LAL1913).19Della Starza I. De Novi L.A. Santoro A. Salemi D. Tam W. Cavalli M. Menale L. Soscia R. Apicella V. Ilari C. Vitale A. Testi A.M. Inghirami G. Chiaretti S. Foà R. Guarini A. Digital droplet PCR and next-generation sequencing refine minimal residual disease monitoring in acute lymphoblastic leukemia. By increasing the number of patients and samples from a single protocol, we can confirm the strength of ddPCR in improving the rate of quantification in critically low positive samples, with a greater concordance compared with NGS (92% versus 87%).In keeping with these observations, in a recent comparative analysis between RQ-PCR and ddPCR performed in a large pediatric cohort in collaboration with the AIEOP cooperative study group,21Della Starza I. Nunes V. Lovisa F. Silvestri D. Cavalli M. Garofalo A. Campeggio M. De Novi L.A. Soscia R. Oggioni C. Mussolin L. Biondi A. Guarini A. Valsecchi M.G. Conter V. Biffi A. Basso G. Foà R. Cazzaniga G. Droplet digital PCR improves IG-/TR-based MRD risk definition in childhood B-cell precursor acute lymphoblastic leukemia. we showed that within a selected subset of childhood ALL patients defined as slow early responders (ie, having a high disease burden at day 33 and resulting RQ-PCR PNQ at day 78), most relapses occurred in cases that proved MRD quantifiable by ddPCR at day 78 (P −4) at day 33 and negative at day 78, and similar to that of medium-risk patients enrolled in the same protocol.Overall, these data indicate that ddPCR is a technique as sensitive as RQ-PCR in detecting and quantifying MRD at all analyzed TPs and more accurate when the RQ-PCR quantitative range is inferior to 10−4, thanks to its greater amplification efficiency and reproducibility. Moreover, ddPCR allows the quantitation of nucleic acid targets without the need of the calibration, thus not limiting the possibility of monitoring patients over time. Several studies have suggested that ddPCR is an attractive tool to monitor MRD in different hematologic malignancies, underlining its capability to predict relapse by quantifying low-positive samples and to lead to a potential refinement in patients' risk stratification.11Cavalli M. De Novi L.A. Della Starza I. Cappelli L.V. Nunes V. Pulsoni A. Del Giudice I. Guarini A. Foà R. Comparative analysis between RQ-PCR and digital droplet PCR of BCL2/IGH gene rearrangement in the peripheral blood and bone marrow of early stage follicular lymphoma.,21Della Starza I. Nunes V. Lovisa F. Silvestri D. Cavalli M. Garofalo A. Campeggio M. De Novi L.A. Soscia R. Oggioni C. Mussolin L. Biondi A. Guarini A. Valsecchi M.G. Conter V. Biffi A. Basso G. Foà R. Cazzaniga G. Droplet digital PCR improves IG-/TR-based MRD risk definition in childhood B-cell precursor acute lymphoblastic leukemia.,23Drandi D. Alcantara M. Benmaad I. Söhlbrandt A. Lhermitte L. Zaccaria G. Ferrante M. Genuardi E. Mantoan B. Villarese P. Cheminant M. Starza I.D. Ciabatti E. Bomben R. Jimenez C. Callanan M. Abdo C. Eckert C. Ribrag V. Cortelazzo S. Dreyling M. Hermine O. Delfau-Larue M.H. Pott C. Ladetto M. Ferrero S. Macintyre E. Droplet digital PCR quantification of mantle cell lymphoma follow-up samples from four prospective trials of the European MCL network.,24Coccaro N. Tota G. Anelli L. Zagaria A. Specchia G. Albano F. Digital PCR: a reliable tool for analyzing and monitoring hematologic malignancies.The new-generation approaches have the remarkable advantage of a greater applicability (≥95% of cases) and sensitivity, and the possibility of proving additional information about the whole clonal immune gene rearrangement status of each patient. At the present time, their use to monitor MRD in clinical protocols is prevented by the lack of published international guidelines, a prerequisite requirement to compare MRD data in different clinical protocols. The EuroMRD Consortium (http://www.euromrd.org, last accessed February 16, 2022) is actively working to rapidly achieve this goal. The next step will be a parallel prospective analysis by ddPCR, for which more data are available, of samples classified as PNQ and/or NEG by RQ-PCR at clinically critical TPs to conclusively clarify its contribution to a further improvement of ALL patients' stratification and outcome.Author ContributionsI.D.S. and L.A.D.N. analyzed the data and wrote the manuscript; A.S., D.S., O.S., M.T., R.S., M.C., V.A., V.B., and E.D.L. performed experiments; F.P. and L.V.C. performed statistical analysis; A.V., M.V., F.F., A.R., and R.B. provided clinical care, collected patient data, and revised the manuscript; and A.G., S.C., and R.F. supervised the experimental work and wrote the manuscript.
Article InfoPublication HistoryPublished online: June 12, 2022
Accepted: April 25, 2022
Publication stageIn Press Journal Pre-ProofFootnotesSupported by Associazione Italiana per la Ricerca sul Cancro , Special 5x1000 Programs “Metastatic Disease: The Key Unmet Need in Oncology” (number 21198 ) (Milan, Italy) (R.F.); and Progetti di Rilevante Interesse Nazionale Italia , 2017PPS2X4 project (S.C.)
I.D.S., L.A.D.N., A.G., and S.C. contributed equally to this work.
Disclosures: None declared.
IdentificationDOI: https://doi.org/10.1016/j.jmoldx.2022.04.014
Copyright© 2022 Association for Molecular Pathology and American Society for Investigative Pathology. Published by Elsevier Inc.
User License Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0) | How you can reuse
留言 (0)