Impact of T-cell Receptor Status on Mutational Landscape and Outcome in T-ALL.

T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive and heterogenous hematological cancer representing 15% of pediatric and 25% of adult ALL.1 Its immunophenotypic classification currently includes early thymic precursor (ETP)-ALLs.2 In 2003, we described and have subsequently used a T-cell receptor (TCR)-based classification to divide T-ALLs into mature sTCRαβ or γδ expressing cases, IMβ/preαβ (complete TCRβ VDJ/cTCRβ expression), and immature IM0/δ/γ cases (cTCRβ negative and no complete TCRβ VDJ).3 Two decades on, we describe a cohort of 426 adult and pediatric T-ALLs patients treated on 2 national prospective protocols (see Suppl. Methods for details), aiming to identify specific oncogenetic subtypes, molecular alterations, clinical presentation, and outcome as a function of their TCR status (IM0/δ/γ; IMβ/preαβ; TCRγδ; TCRαβ). Two cohorts were included: 220 children, treated according to the FRALLE2000T recommendations, and 206 adults, treated in the GRAALL03/05 protocol. Overall, the IMβ/preαβ immunophenotype predominated (51%, n = 215) followed by IM0/δ/γ (22%, n = 92), TCRαβ (15%, n = 66), and TCRγδ (12%, n = 53) (Table 1). Table 1 - Biological and Clinical Characteristics of GRAALL03/05 and FRALLE2000T-treated Patients According to the TCR Subgroup All Cohorts IM0/δ/γ; N = 92 (22%) IMβ/preαβ; N = 215 (51%) TCR αβ; N = 66 (15%) TCR γδ; N = 53 (12%) Total; N = 426 P-Value Genotype subsets PICALM::MLLT10 5 (5%) 2 (1%) 2 (3%) 4 (8%) 13 (3%) SET-NUP214 2 (2%) 1 (0.4%) 0 1 (2%) 4 (1%) SIL-TAL1 0 29 (13%) 23 (35%) 0 52 (12%) TLX1 2 (2%) 49 (23%) 1 3 (6%) 55 (13%) TLX3 3 (3%) 44 (20%) 3 (5%) 17 (32%) 67 (16%) HOXA9 overexpression 31/69 (45%) 23/163 (14%) 4/44 (10%) 18/44 (41%) 76/320 (24%) ETP 42/66 (64%) 9/161 (6%) 1/42 (2%) 4/38 (11%) 56/307 (18%) Oncogenetic classifier NOTCH1/FBXW7 mut 56/91 (62%) 169 (79%) 31 (47%) 35 (66%) 291/425 (68%) RAS mut /PTEN alt 20/91 (22%) 44 (20%) 25 (38%) 10 (19%) 99/424 (23%) High-risk classifier a 46/91 (51%) 73 (34%) 41 (62%) 24 (45%) 184/425 (43%) Clinical subsets Age, median (range), y 23.7 (1.2–59.1) 16.3 (1.1–57.2) 13.3 (2.4–49.9) 14.8 (1.3–59) 16.2 (1.1–59.1) <0.001 WBC, median (range), G/L 30.8 (0.3–710) 61 (3.2–965) 119 (2.2–980) 103 (3.8–645) 63.3 (0.3–980) <0.001 CNS involvement (%) 10/89 (11%) 22/211 (10%) 11/66 (17%) 2/53 (4%) 45/419 (11%) 0.16 Male, n (%) 61 (66%) 160 (74%) 51 (77%) 45 (85%) 317 (74%) 0.09 Treatment response Good early prednisone response 38 (41%) 140/211 (66%) 29/63 (46%) 23 (43%) 230/419 (55%) <0.001 CR 83 (90%) 203 (94%) 63 (95%) 45 (85%) 394 (92%) 0.07 MRD >10−4 35/52 (67%) 36/160 (22%) 17/49 (35%) 20/41 (49%) 108/302 (36%) <0.001 Allo HSCT 34/86 (39%) 33/205 (16%) 11/64 (17%) 16/49 (33%) 94/404 (23%) <0.001 aOncogenetic classifier was performed for 425 patients.P-values were evaluated by χ2 test. GPR was defined as <1000 circulating blasts/µL on day 8. MRD was evaluated at D35. Notch/FBXW7-RAS/PTEN classifier (oncogenetic classifier) was performed as previously described.4Allo HSCT = allogeneic hematopoietic stem cell transplantation in 1st CR; alt = altered; CNS = central nervous system; CR = complete remission; ETP = early thymic precursor; GPR = good prednisone response; MRD = minimal residual disease; mut = mutated; WBC = white blood cells. TCR subtype by age (Figure 1A) showed that IM0/δ/γ and TCRγδ are relatively common in older patients, whereas αβ-lineage (IMβ/preαβ and TCRαβ) T-ALLs predominate in patients <25 years (Suppl. Figure S1). ETP-ALLs were, as expected, most common in IM0/δ/γ (64%) but were also identified in 11% of TCRγδ T-ALLs compared with only 4.9% (10/203) of αβ-lineage T-ALLs (Table 1).4 As such, ETP-ALLs are not synonymous with immunogenetic immaturity, but rarely undergo complete TCRβ rearrangement.Figure 1.: Distribution and outcome of T-ALL according to the TCR subgroup. (A) Proportion of T-ALL with different TCR subgroups according to the age (n = 426 samples). Patients aged from 0 to 14 y: n = 198; from 15 to 24 y: n = 87; from 25 to 39 y: n = 86; from 40 to 59 y: n = 55. (B) Overall survival of patients according to the TCR subgroup (FRAALLE + GRAALL) (left). Cumulative incidence of relapse of patients according to the TCR subgroup (FRALLE + GRAALL) (right). (C) Overall survival of patients TCRαβ (left) and of patients non-TCRαβ according to the protocol (right). (D) Cumulative incidence of relapse of patients TCRαβ (left) and of patients non-TCRαβ according to the protocol (right). T-ALL = T-cell acute lymphoblastic leukemia; TCR = T-cell receptor.Oncogenetic characteristics according to the TCR subtype are described in Table 1 and Suppl. Tables S1 and S2 for adults and children, respectively. The rare PICALM-MLLT10 predominated in IM0/δ/γ and TCRγδ T-ALLs (9/145; 6.2% versus 4/281; 1.4% in αβ-lineage T-ALLs, P = 0.01). HOXA9 was overexpressed in 24% of T-ALL, predominantly in IM0/δ/γ (45%) and TCRγδ (41%).5TLX1 T-ALLs predominated in the IMβ/preαβ subgroup (23%) and TLX3 in TCRγδ cases (32%). SIL-TAL1 was most common in TCRαβ T-ALLs and totally absent in IM0/δ/γ and TCRγδ T-ALLs.6 RNA sequencing was performed in a subgroup of 112 T-ALL (92 GRAALL03/05 and 20 FRALLE2000T) with available RNA. Uniform manifold approximation and projection for dimension reduction (UMAP) distribution is represented in Suppl. Figure S2. The IM0/δ/γ subgroup represented a homogeneous cluster, co-segregating with ETP status and HOXA9 overexpression for some, whereas the IMβ/preαβ subgroup split into 2 clusters: approximately half cosegregated with the SIL-TAL1/TCRαβ cluster; the other half with TLX1/TLX3 overexpression. The TCRαβ subgroup was more homogeneous than TCRγδ cases, which were transcriptionally close to either the IM0/δ/γ subgroup or within the IMβ/preαβ/TLX1/3 subgroup. The mutational landscape of the overall cohort is represented in Suppl. Figure S3A and Suppl. Table S3. The overall median number of mutations-deletions/patient was 5 (range, 0–21) and was lower in the TCRαβ subgroup compared with others, with a median of 4 alterations/patient (range, 1–10), P < 0.001 (Suppl. Figure S3B). In IM0/δ/γ T-ALLs, more mutations were identified in the SUZ12 (41%) and EZH2 (19%) polycomb repressive complex 2 compared with αβ-lineage T-ALLs. JAK/STAT signaling gene mutations, particularly JAK3 (44%) and JAK1 (17%), were also relatively common, whereas PI3KR1 and PTEN alterations were less common than in TCRαβ T-ALLs, in keeping with the mutual exclusion of these 2 signaling pathways in T-ALLs. More NRAS (21%) and NF1 (24%) mutations (RAS signaling) and DNMT3A (16%) and IDH2 (7%) mutations (epigenetic factors) were identified, whereas CDKN2A/B cell cycle alterations were relatively uncommon (26%). In IMβ/preαβ T-ALLs, BCL11B (31%) and LEF1 (18%) mutations (transcription factors) and PTPN2 (12%) mutations were relatively common and almost all had NOTCH1 (88%) and/or FBXW7 mutations (38%) and CDKN2A/B alterations (84%). TCRαβ T-ALLs presented predominantly with NOTCH1 mutations (63%) and PTEN (57%) and CDKN2A/B alterations (87%). Relatively few mutations in IL7/JAK/STAT pathway, epigenetic factors, and RAS signaling were found, including compared with IMβ/preαβ T-ALLs for the latter 2 categories. TCRγδ T-ALLs demonstrated a mutational landscape more similar to IM0/δ/γ than to TCRαβ T-ALLs, with relatively frequent mutations in RAS signaling genes (NF1 23%, NRAS 9%), epigenetic factors, and RUNX1 (17%). The clinico-biological characteristics of GRAALL03/05 and FRALLE2000T treated patients are described in Table 1 and Suppl. Tables S4 and S5. Although clinical presentation (age and white blood cell [WBC] count) and early response to prednisone (EPR) differed between TCR subtypes, this had little impact on complete remission rates, which were >90% in all but TCRγδ T-ALLs (85%; 87% in children and 83% in adults). Many characteristics were similar in children and adults, with the only major overall differences being WBC at presentation (median 105 G/L and 36.6 G/L, respectively) and use of allogeneic hematopoietic stem cell transplantation (Allo HSCT) (10% versus 37%). Patients with IM0/δ/γ T-ALL presented with a relatively low WBC (median 30.8G/L). They had a poorer EPR (41%), more frequent minimal residual disease (MRD) positivity at the end of induction (EoI) (67%; 61% in children and 74% in adults) and 51% were high risk with the NOTCH1/FBXW7/RAS/PTEN (N/F/R/P) oncogenetic classifier. They underwent Allo HSCT in first complete remission relatively frequently (39%). In IMβ/preαβ patients, a good EPR was obtained for 66%, with less frequent EoI MRD positivity, particularly in adults (22%; 29% in children and 12% in adults), and less Allo HSCT (16%) than in the overall cohort. Oncogenetic high-risk status was relatively rare in this category (34%). Patients with TCRαβ T-ALL presented with high hyperleukocytosis (median WBC = 119 G/L) and relatively frequent CNS involvement (17%). Despite a poor EPR in most patients from both the age groups, EoI MRD positivity was only 35% overall and Allo HSCT was not performed more frequently (17%). In keeping with the high incidence of PTEN abnormalities, many demonstrated an oncogenetic high-risk classifier (62%). Patients with TCR γδ T-ALL presented with high WBC (median WBC = 103 G/L) but rare CNS involvement (4%). There were few good EPR (43%) and frequent EoI MRD positivity (49%). Allo HSCT in first line was undertaken in 33% overall, when it was disproportionately common in children (27%). Long-term 5-year overall survival (OS) and cumulative incidence of relapse (CIR) are represented in Figure 1B and Suppl. Table S6 for the entire cohort and per protocol (Suppl. Figure S4, Suppl. Figure S5). There was ~9% superiority in 5-year OS and a 5% lower CIR rate in children when compared with adults. Despite different clinical presentations and initial therapeutic response profiles, no significant difference in long-term outcomes could be identified among the different TCR subtypes. Relapses in TCRαβ T-ALLs occurred earlier, with few occurring after 1.5 years. When 5-year OS and CIR was compared between protocols (Suppl. Table S6 and Figure 1C–D, Suppl. Figure S4), no significant differences were seen between the 4 TCR subgroups individually, although there was a trend for superior 5-year OS in the FRALLE versus GRAALL-treated IMβ/preαβ patients (P = 0.07, Suppl. Table S6). This trend was also seen for all non-TCRαβ T-ALLs but was absent in the TCRαβ subgroup. Comparison of outcome by protocol for TCRαβ versus non-TCRαβ showed significantly higher 5-year OS (P = 0.009) in children compared with adults (Figure 1C), but not for TCRαβ T-ALLs (P = 0.58), when children did as badly as adults (5-year OS 69% in both groups). No significant differences were seen in the 5-year CIR rates in TCRαβ or non-TCRαβ T-ALLs (Figure 1D), suggesting that something in addition to early relapse contributes to the poor outcome of children with TCRαβ T-ALL treated according to the FRALLE2000T recommendations. Classification of T-ALL has evolved from immunophenotypic identification of lineage to genetic identification of oncogenetic subtypes, which often correlate with both stage of maturation arrest and clinical outcome.7–13 Immunogenetic characterization of TCR (and immunoglobulin) gene rearrangements is rarely required for diagnosis, but is widely used for MRD evaluation.14–16 TCRαβ T-ALLs fared particularly badly (5-year OS 68.5%) in FRALLE-treated patients, a protocol that was shown to be suboptimal for pediatric oncogenetic high-risk patients.17 In fact, 5-year OS in the present study was virtually identical in pediatric and adult TCRαβ patients, despite the lower 5-year CIR in children. Possible explanations may include the relatively high incidence of CNS involvement and less frequent use of Allo HSCT. Pediatric TCRαβ T-ALLs clearly need alternative first-line therapy. Some are intensified on the basis of frequent high leukocytosis, but inclusion of TCRαβ expression in stratifying characteristics merits consideration. Defining oncogenic features of IM0/δ/γ T-ALLs overlap with those described for ETP-ALLs in children and adults,2,5 although only 64% were phenotypically defined as such. Adult IM0/δ/γ T-ALLs benefit from Allo HSCT, but still trend to inferior survival, despite CIR rates comparable to non-TCRαβ T-ALLs.5 In contrast, although only 12% of FRALLE2000T-treated IM0/δ/γ T-ALLs underwent Allo HSCT, 5-year OS was comparable to other pediatric TCR subgroups. IM0/δ/γ T-ALL profiles and response to treatment also overlapped with some TCRγδ cases. Both commonly overexpressed HOXA9, which identifies a poor prognostic group in immature GRAALL-treated T-ALLs.5 Further understanding of TCRγδ T-ALLs requires more in-depth characterization of the 2 clusters identified here. The lack of overall prognostic significance of this TCR classification limits its use in individual stratification. Our data demonstrate, however, that integration of this parameter alongside oncogenetic annotations adds understanding, encouraging its exploitation in immunogenetic MRD-driven protocols. At a minimum, immunophenotypic diagnostic panels should specify lineage in sTCR positive T-ALLs of all ages. There were many more similarities than differences between pediatric and adult T-ALLs, arguing for concerted evaluation and common protocols. The data presented here should contribute to selecting subgroups of T-ALL, which may respond differently to specific therapeutic regimens, as recently described for the N/F/R/P oncogenetic classifier in the FRALLE2000T and UKALL2003 pediatric regimens.17 Such reversed use of discordant responses within oncogenetically or immunogenetically/TCR-defined subgroups will throw light on the fine details of therapeutic practice in ALL. ACKNOWLEDGMENTS The authors thank all participants in the GRAALL-2003 and GRAALL-2005 study groups, the SFCE, and the investigators of the 16 SFCE centers involved in collection and provision of data and patient samples, and V. Lheritier for collection of clinical data. AUTHOR CONTRIBUTIONS MED, EM, and VA conceived the study and oversaw the project. MED, LC, MS, AB, NB, AP, AT, LL, and VA provided study materials or patients. MED and GA performed molecular analyses. MED, LC, VA, and GA collected and assembled data. MED and GA performed statistical analysis. MED, GA, and VA analyzed and interpreted data. MED, EM, and VA wrote the article. All authors approved the article. DATA AVAILABILITY This work did not generate new data. DISCLOSURES EM reports that she is president of the European Hematology Association. All the other authors have no conflicts of interest to disclose. SOURCES OF FUNDING The GRAALL was supported by grants P0200701 and P030425/AOM03081 from the Programme Hospitalier de Recherche Clinique, Ministère de l’Emploi et de la Solidarité, France and the Swiss Federal Government in Switzerland. Samples were collected and processed by the AP-HP “Direction de Recherche Clinique” Tumor Bank at Necker-Enfants Malades. MED was supported by “CARPEM.” MS was supported by “Action Leucémie” and “Soutien pour la formation à la recherche translationnelle en cancérologie dans le cadre du Plan cancer 2009–2013”. LC was supported by “Fondation pour la Recherche Médicale.” GA was supported by “Fondation de France.” This work was supported by grants to Necker laboratory from the “Association Laurette Fugain,” Institut National du Cancer PRT-K 18-071 and PLBIO2021-097-PL. All the other authors have no conflicts of interest to disclose.