Page 60 - 2021_06-Haematologica-web

S. Chiaretti et al.
Figure 1. Distribution of the genetic lesions in the Philadelphia-like (Ph-like) and non-Ph-like cases study. Only the samples evaluated for the BCR/ABL1-like pre- dictor and mutational status are depicted.
hotspot R683. IL7R and CRLF2 were mutated in two sam- ples, while JAK3 only in one. Furthermore, six of the eight mutated samples (75%) displayed a concomitant CRLF2 overexpression. Nine RAS pathway mutations - only one being clonal - were found in six patients (25%). The most frequent mutations (n=5) involved the hotspot G12-13 of KRAS and NRAS. CNA analysis in Ph-like cases revealed IKZF1, BTG1, CDKN2A/2B, PAX5 and EBF1 deletions in 14 (63.6%), 11 (50%), seven (31.8%), seven (31.8%) and six (27.3%) cases, respectively. Furthermore, IKZF1 + CDKN2A/2B and PAX5 deletions, known to confer a very poor outcome, were identified in 10 cases (45.5%). Finally, RNA-sequencing and/or FISH experiments of the Ph-like ALL cases revealed 11 TK activating lesions (47.8%): five ABL-class fusion genes (three NUP214/ABL1, one ZC3HAV1/ABL2 and one EBF1/PDGFRB), two BCR/JAK2, three CRLF2-r and 1 DDX3X/USP9X, the latter known to be associated with CRLF2 deregulation.26
Overall, Ph-like associated lesions were identified in 70.8% (17 of 24) of cases and are summarized in Table 2. When the genetic landscape of Ph-like ALL was com- pared to that of the non-Ph-like cases, significant differ- ences emerged. As shown in Table 3, CRLF2-high was significantly more frequent in Ph-like ALL (35.7% vs. 13.3%, P=0.018). Similarly, clonal JAK/STAT mutations were specific of the Ph-like subset (33.3% vs. 4%, P=0.001), while RAS pathway clonal mutations were more frequent in non-Ph-like than in Ph-like ALL cases (46% vs. 4.2%, P=0.001). Coincidence analysis (CNA) documented that IKZF1, EBF1 and BTG1 deletions were significantly more common of the Ph-like than in the non-Ph-like subset (63.6%, 50% and 27.3% vs. 25%, 7.8% and 2.1%, respectively; P=0.002, P<0.001 and P=0.007); CDKN2A/2B and PAX5 deletions were equally distributed among Ph-like and non-Ph-like cases (31.8%
vs. 47.9% and 31.8% vs. 22.9%, respectively).
The analysis of fusion genes, performed on a total of 85 patients, showed that rearrangements involving TK or cytokine receptors were significantly higher in the Ph-like cases with ten fusion genes involving either CRLF2 or a TK compared to only one CRLF2-r case in the non-
BCR/ABL1-like cases (43.5% vs. 1.6%, P<0.001).
The genetic lesions documented in both the Ph-like and
non-Ph-like subgroups are detailed in the Online Supplementary Table S3 and their distribution is provided in Figure 1; further details on non-Ph-like ALL cases, as well as on NGS coverage, are provided in the Online Supplementary Results and Online Supplementary Table S5, respectively.
Response to treatment, minimal residual disease evaluation and transplant allocation
The Ph-like status was significantly associated with response to treatment: in fact, Ph-like patients had a sig- nificantly inferior CR rate at time point 1 (TP1) compared to non-Ph-like cases (74.1% vs. 91.5%, P=0.044, Table 4) and this translated into a lower probability of CR achievement (P=0.038, OR=0.265, 95% Confidence Interval [CI]: 0.071-0.921, Online Supplementary Table S6). The latter data retained statistical significance also in a multivariate model adjusted for clinically relevant param- eters, as well as for genetic lesions with a prognostic rel- evance.
MRD evaluation - feasible in 64 patients at TP1, 62 at TP2 and 49 at TP3 - showed that at TP1, 77.8% of Ph-like cases and 41.3% of non-Ph-like were MRD-positive (P=0.012); at TP2, 52.9% of Ph-like cases and 20% of non-Ph-like were MRD-positive (P=0.025); similarly, at TP3, 41.7% of Ph-like cases and 13.5% of non-Ph-like cases were MRD-positive (P=0.05). These data, summa- rized in Table 4, indicate that in the Ph-like patients there is a significantly higher MRD persistence at all TP evalu- ated compared to non-Ph-like cases. Consistently, the univariate analyses for MRD results showed that - when considering both clinically relevant parameters and genet- ic prognostic markers - only the Ph-like status was a risk factor for being MRD-positive at TP2 (P=0.014, OR=4.5, 95% CI: 1.373-15.508) (Table 5).
As a consequence, HSCT rate in first CR was signifi- cantly higher (P=0.015) in Ph-like vs. non-Ph-like cases (eight of 20 vs. 6 of 54, 40% vs. 11%, respectively), in line with the guidelines of the trial, in which MRD persist- ence was a criterion for HSCT allocation. Importantly, among five MRD+ Ph-like patients who did not undergo a transplant, four relapsed at a median period a 7.8 months from CR, whereas no relapses occurred in the three MRD+ Ph-like patients undergoing HSCT.
1562
haematologica | 2021; 106(6)