Somatic co‐alteration signatures are prognostic in high‐grade TP53 ‐mutated myeloid neoplasms

Summary

To assess the relevance of co‐occurring somatic mutations in TP53‐mutated myeloid neoplasms with ≥10% blasts, we pooled 325 individuals from 10 centres. We focused on comparing three published somatic co‐alteration signatures comprising (1) nine MDS‐related genes (‘ICC‐MDSR’), (2) ICC‐MDSR + additional secondary mutations‐related genes (‘Tazi signature’) and (3) EPI6 (comprising six genes). Outcomes examined were 24‐month overall survival (OS24) and front‐line complete response (CR1). The median age was 69 years with 77% receiving front‐line hypomethylating agents (HMA). All three signatures ICC‐MDSR (p = 0.009), Tazi signature (p = 0.001) and EPI6 (p = 0.025) predicted inferior CR1. In the low‐intensity (HMA) subgroup, only Tazi signature (p = 0.026) predicted inferior CR1. In OS24 analysis of the HMA‐treated subgroup (N = 200), only Tazi signature was adverse (hazard ratio, HR = 1.6 [1.1–2.2]; p = 0.011). However, a forward stepwise multivariable age‐adjusted Cox model including all three signatures picked EPI6 as the sole significant adverse predictor in the entire cohort (p = 0.0001) as well as within the HMA‐treated subgroup (p = 0.0071). These data confirm the value of testing co‐occurring somatic alterations even within a high‐grade TP53‐mutated myeloid neoplasm cohort.

INTRODUCTION

The World Health Organization 5th (WHO5) edition classification of haematopoietic neoplasms recognizes eight genes as ‘secondary mutations‐related genes’ (SRSF2, SF3B1, U2AF1, ZRSR2, ASXL1, EZH2, BCOR and STAG2) with RUNX1 additionally recognized by the ICC. ¹ The International Consensus classification (ICC)‐defined acute myeloid leukemia (AML) and Myelodysplastic syndrome (MDS)/AML with MDS‐related [ICC‐MDSR] gene mutations category explicitly exclude TP53 from this list given the overarching adverse prognostic effect of a TP53 alteration ¹ , ² mandating a unique category for TP53‐altered myeloid neoplasms (MNs). Regardless, both entities are considered high risk as per ELN22 risk stratification for MNs. ³ These data align with the study by Tazi and colleagues ⁴ designating secondary AML as a distinct class characterized by ‘secondary’‐type mutations (SMs) but lacking TP53 mutations. TP53 mutations were enriched in the ‘Complex Karyotype’ class in the study. ⁴

Recently, we described another novel EPI6 co‐alteration signature comprising only six genes (CUX1, CBL, EZH2, KRAS, TET2 and U2AF1) with predictive ability for outcomes within TP53 ^MUT MNs. ⁵ In this study, CUX1‐altered patients in particular experienced very poor front‐line response as well as 24‐month survival. With this background, we evaluated all 3 published co‐mutation signatures, the 14‐gene Tazi et al. signature, ⁴ the 9‐gene ICC‐MDSR ¹ as well as the recently described 6‐gene EPI6 ⁵ in this expansion cohort comprising 324 individuals to see which of these hold up for predicting key outcomes.

MATERIALS AND METHODS

Cohort case selection and sample procurement

Individuals with high‐grade TP53 ^MUT MN (≥10% blasts) carrying ≥1 TP53 mutation at a variant allele frequency (VAF) ≥3% diagnosed between 2014 and 2024 were considered for inclusion. All pertinent baseline pathology, laboratory and clinical data were abstracted. Response was assessed per ELN 2017 guidelines ⁶ denoting complete remission (CR), CR with incomplete haematological recovery (CRi) and morphological leukaemia‐free state (MLFS) as composite measures of front‐line complete response (CR1).

Somatic next‐generation sequencing (NGS) was performed per each institution's clinical grade panel at the time of diagnosis. Pathogenic somatic mutations in genes relevant to all three signatures were assessed (except MLL‐PTD) (See Supplementary Section 1; Figure S1, Venn diagram for unique and overlapping genes). Since exploratory analysis revealed no significant differences in outcomes between 1 and 2 + mutation in the Tazi signature, we denoted 1 or more co‐alteration/mutation as being abnormal for all three signatures. Of note, copy number losses by NGS were also marked as equivalent to a mutation (especially with CUX1 and NF1). TP53 allelic state also was assessed per ICC 2022¹ but single‐hit (TP53 ^SH ) designation used an expanded VAF cut‐off including cases with VAF between 3% and 10%.

Statistical analysis

Outcomes assessed were 24‐month overall survival (OS24) from the time of diagnosis of TP53 ^MUT MN and response to first‐line therapy. Data were analysed using Kaplan–Meier, Cox proportional hazards (P‐H) regression ⁷ or flexible parametric models with P‐H assumption violation. ⁸ (Supplementary Section 2).

RESULTS

Table 1 summarizes the baseline characteristics and therapy/response data of all patients stratified by Tazi signature at diagnosis of TP53 ^MUT MN. At baseline, most had complex karyotypes (79%; 231/293) with several recurrent structural alterations in 94% of patients. Tazi signature was marginally associated with TP53 ^MH (p = 0.07). As expected, ICC‐MDSR (p < 0.0001) and EPI6 signatures (p < 0.0001) correlated significantly with Tazi signature due to overlapping genes (Table 1).

TABLE 1.

Baseline characteristics of all 325 patients at diagnosis of TP53mutt myeloid neoplasm stratified by Tazi et al ⁴ signature.

	Score 0	Score 1+	Test
	N = 222 (68.3%)	N = 103 (31.7%)
Age at diagnosis, median [IQR]
Age (years)	67.9 [14–93]	70.4 [25–92]	0.012
Baseline labs, median [IQR]
Haemoglobin (g/dL)	8.2 [3–13]	8.1 [4–17]	0.810
Platelet count (103/μL)	43.0 [3–584]	45.0 [6–585]	0.117
Abs. Neut. count (103/μL)	0.6 [0–100]	0.8 [0–27]	0.213
Sex
Male	118 (53.2%)	62 (60.2%)	0.235
Female	104 (46.8%)	41 (39.8%)
Prior Myeloid Neoplasm a —no. (%)
No	181 (81.5%)	77 (74.8%)	0.160
Yes	41 (18.5%)	26 (25.2%)
MDS/AML‐ICC 2022
MDS/AML (10%–19% blasts)	48 (21.6%)	19 (18.4%)	0.510
AML (20% + blasts)	174 (78.4%)	84 (81.6%)
−5/5q loss
Absent	65 (32.0%)	35 (38.9%)	0.253
Present	138 (68.0%)	55 (61.1%)
−7/7q loss
Absent	100 (48.1%)	45 (48.4%)	0.960
Present	108 (51.9%)	48 (51.6%)
Chrom. 17/17p deletion
Absent	127 (62.6%)	48 (53.3%)	0.137
Present	76 (37.4%)	42 (46.7%)
Highest TP53 VAF %
≤25%	49 (22.2%)	15 (14.6%)	0.109
>25%	172 (77.8%)	88 (85.4%)
MDS‐related gene mutn.
Absent	222 (100.0%)	28 (27.2%)	<0.001
Present	0 (0.0%)	75 (72.8%)
EPI6 signature
Absent EPI6	206 (92.8%)	45 (43.7%)	<0.001
Present EPI6	16 (7.2%)	58 (56.3%)
Therapies—no. (%)
Low‐inten. (HMA ± VEN)	144 (67.3%)	65 (65.0%)	0.878
Intensive (Chemo)	43 (20.1%)	19 (19.0%)
Best.Supp.Care	19 (8.9%)	11 (11.0%)
CPX‐351/Vyxeos	8 (3.7%)	5 (5.0%)
HMA therapy group
HMA ± placebo	74 (52.5%)	21 (33.3%)	0.011
HMA + Ven	67 (47.5%)	42 (66.7%)
CRc b —no. (%)
No CR/CRi/MLFS	110 (65.1%)	71 (84.5%)	0.001
CR/CRi/MLFS	59 (34.9%)	13 (15.5%)
Allo‐SCT—no. (%)
No AlloSCT	179 (80.6%)	95 (92.2%)	0.007
Allo‐SCT	43 (19.4%)	8 (7.8%)

Abbreviations: BSC, best supportive care; CR, complete response; CRi, CR & incomplete haematological recovery; IQR, interquartile range; Med., median; MLFS, morphological leukaemia‐free state.

^a These included treated or untreated MDS (Low‐ and High‐risk), MDS/MPN and MPN without any TP53mutt up until evolution/progression to a TP53mutt myeloid neoplasm.

^b Numbers reported in therapies and response may not add up to cohort total since some patients were either untreated or if treated; response could not be evaluated due to various reasons (active second malignancy, early treatment‐emergent adverse effects, transferred care elsewhere or early mortality).

Somatic NGS uncovered 400 pathogenic mutations in TP53 (242 unique mutations) with hot‐spot DNA‐binding domain missense mutations predominating. Co‐alterations were observed in 58% (N = 188) with a median of one co‐alteration (range: 0–16 co‐alterations). Co‐alterations in epigenetic pathway (DNMT3A and TET2) genes predominated, in line with prior observations. ⁹ All three signatures were associated with significantly higher TP53 VAF (Figure S2 violin plots).

First‐line treatment and response information

Among the 314 patients with available treatment information, low‐intensity regimens, mostly hypomethylating agents (HMA) based, were used in 67% (N = 209), intensive chemotherapy in 20% (N = 62), CPX‐351 in 4% (N = 13) while 10% (N = 30) received only best supportive care. First‐line response was evaluable in 253 patients with 28% achieving CR, 25% having partial response and 47% having non‐response/stable‐progressive disease.

Predictors of response

Looking at the entire cohort, male gender (23% vs. 36% CR1 in females; p = 0.027), ICC‐MDSR (16% vs. 33%; p = 0.009), EPI6 (16% vs. 32%; p = 0.025) as well as Tazi signature (15% vs. 35%; p = 0.001) predicted inferior front‐line response. In the low‐intensity subgroup (HMA ± Venetoclax [VEN]), only Tazi signature predicted inferior front‐line response (21% vs. 38% CR1 in Score 0; p = 0.026). Among the 95/204 receiving HMA only (without VEN), Tazi signature (p = 0.07) predicted inferior response, but not ICC‐MDSR (p = 0.15) or EPI6 (p = 0.31). A total of 17% (51/295) went on to receive subsequent allogeneic stem cell transplantation (excluding individuals receiving supportive care only).

Baseline outcome data

Entire cohort analysis

The median duration of follow‐up (from diagnosis of TP53 ^MUT MN to study exit) was 6 months (range: 0–73 months) with a 24‐month survival of 14% (95% CI = 10.2%–18.6%). Age at diagnosis ≥70 years (HR = 1.5 [1.2–1.9]; p = 0.001) and complex karyotype (HR = 1.9 [1.3–2.7]; p < 0.001) predicted worse OS24. We next assessed the relevance of these signatures in the entire cohort as well as subsets by age and therapy where Tazi signature remained relevant in the entire cohort, more so in older patients as well in the low‐intensity subgroup (Figure 1A,B). In additional subset analysis, Tazi signature was relevant within the subgroup harbouring isolated TP53 ^MIS mutations (HR = 1.5 [1.1–2.1]; p = 0.007; N = 230) and the subgroup with TP53 ^SH allelic state (HR = 1.5 [1.1–2.0]; p = 0.009; N = 245).

FIGURE 1.

Impact of co‐mutation signatures on 24‐month survival. (A) Entire cohort analysis showing significant adverse impact with Tazi signature on 24‐month survival. (B) This adverse impact was even more pronounced in the subgroup of older patients. (C) Among HMA‐treated (HMA+/−Ven) patients, Tazi signature retained its adverse significance. Although not shown, in a subgroup analysis of patients treated only with HMA (no venetoclax) landmarked at 2 months after diagnosis excluding patients with early mortality, Tazi signature predicted inferior outcomes with no survivors at 24 months in the Tazi score 1+ group. (D) Likewise, EPI6 also significantly impacted outcomes in the low‐intensity subgroup. ICC‐MDSR was, however, not relevant for outcomes in subgroup analysis. Median survival and 24‐month OS% depicted by group. [Colour figure can be viewed at wileyonlinelibrary.com]

Low‐intensity subgroup analysis

In analysis restricted to HMA‐treated individuals, HMA + VEN offered no OS24 benefit (HR = 1.1 [0.7–1.8]; p = 0.58; N = 107) compared to HMA only. Within the HMA+/‐VEN subgroup, Tazi score had a significant adverse impact (HR = 1.6 [1.1–2.2]; p = 0.011; N = 200; Figure 1C) besides EPI6 (Figure 1D) while ICC‐MDSR was not relevant. Also, see Figure S3 for sensitivity analysis excluding individuals with TP53 mutations with VAF < 10%.

Multivariable analysis

We conducted a forward stepwise multivariable Cox regression with all three co‐mutation signatures (ICC‐MDSR, Tazi and EPI6) adjusted for age at diagnosis. In this analysis, only the parsimonious EPI6 remained as the sole significant adverse predictor of OS24 in the entire cohort (p = 0.0001) as well as in the low‐intensity subgroup (p = 0.0071). Achieving CR1 to first‐line significantly improved outcomes (OS24 34.8% vs. 6.8% for no CR1; p < 0.0001) overall and within treatment subgroups of intensively treated or HMA‐treated individuals (OS24 27.7% vs. 3.6% for no CR1; p < 0.0001).

Among transplanted individuals, only EPI6 at diagnosis showed a trend towards predicting relapse at day + 100 post‐allogeneic stem cell transplantation (alloSCT; p = 0.07) while neither ICC‐MDSR (p = 0.74) nor Tazi score (p = 0.69) predicted relapse by day 100 or OS24. Only del(17p) at diagnosis predicted worse survival among transplanted individuals (14.7 vs. not reached months; p _Log‐rank  = 0.004). Lastly, we constructed another multivariable Cox model including pretherapy (EPI6, TP53 VAF) and therapy predictors (CR1 and transplant) stratified by age where EPI6 retained its significance with an impressive model Harrell's concordance statistic of 0.71.

DISCUSSION

Although AML and MDS/AML with either TP53 mutations or MDS‐related gene mutations are recognized to be mutually exclusive categories per ICC designation, ¹ our study underscores the value of evaluating additional non‐TP53 co‐mutations even within an ELN22 high‐risk TP53 ^MUT cohort. While none of the three somatic co‐alteration signatures was relevant for predicting response in the low‐intensity subgroup (HMA ± VEN), in the subgroup analysis of HMA‐only patients, the Tazi signature retained its predictive value for inferior front‐line response. In looking for genes within this signature most predictive of inferior response, we recently identified that CUX1 alterations predict very poor front‐line response in this cohort. ⁵ In light of these data, it seems plausible that CUX1 alterations (included in the Tazi as well as the EPI6 signatures) may account for the poor response in this expansion cohort with the Tazi signature although the underlying mechanistic explanation for this remains unclear.

Intriguingly, a greater proportion of patients harbouring the Tazi signature received VEN‐based therapies (66% vs. 44%) despite largely similar blast counts between the two groups. However, neither blast counts (10%–19% vs. 20%+) nor addition of VEN separately impacted survival, indicating that the impact of the Tazi signature is independent of blast counts or choice of therapy. These data are in keeping with recent data reporting little added benefit with VEN, especially with TP53 ^MUT MN. ¹⁰ , ¹¹ Comparing all three signatures (14‐gene Tazi, 9‐gene ICC‐MDSR and 6‐gene EPI6), we found that the parsimonious 6‐gene EPI6 is sufficient for explaining the variance in outcomes in this high‐risk cohort with over 300 patients—a major strength lending power to all analysis. In all fairness, however, both ICC‐MDSR and Tazi signatures were developed outside the context of TP53‐mutated disease, whereas EPI6 was specifically designed within a TP53‐mutated cohort. In our observations, the dominant effect within the EPI6 signature seems to be driven by CUX1, with deletions at the CUX1 locus—often associated with del(7q)—being much more prevalent than mutations, which were more commonly seen in the Tazi cohort. Whether CUX1 deletions play a cooperative biological role in TP53‐mutated neoplasms or are merely an epiphenomenon of del(7q), remains an open question. Nevertheless, our data provide a strong rationale for routinely incorporating CUX1 into NGS panels, alongside copy number alteration analysis (for CUX1, EZH2 and TET2).

A couple of drawbacks merit consideration. First, we have not adjusted these models for prior HMA exposure in the setting of an antecedent MN. Second, all these signatures are VAF agnostic for the co‐mutated gene making it hard to decipher if some of the observed co‐mutations at a low VAF merely represent background clonal haematopoiesis unrelated to the leukaemic clone. While the parsimonious EPI6 signature explains the most variance in outcomes, it would be important to replicate these results in a VAF‐weighted co‐mutation signature considering only co‐mutations that are present at a VAF ≥10% VAF which is less likely to be CHIP related.

In summary, our data support testing of co‐mutations even in high‐grade TP53 ^MUT MN with the parsimonious EPI6 emerging as the most relevant signature. Whether EPI6 has any value in predicting outcomes in MDS/AML or AML outside the context of TP53‐mutated disease will remain to be seen. As the treatment landscape evolves for this high‐risk cohort, these data will better inform the right therapeutic strategy for this subgroup with co‐mutations since these individuals are very unlikely to achieve a CR, receive an allogeneic stem cell transplant or derive durable post‐transplant remission even if transplanted.

AUTHOR CONTRIBUTIONS

Study conception and design: P.W. and G.V.; Provision of study materials: E.S., P.W., P.S., M.M., A.A.P., F.H., S.G., G.R.R., Q.Z., A.F., T.B., J.Z., H.T., H.C., R.C.B., A.P. and G.V.; Data collection and assembly: E.S. and G.V.; Data analysis and interpretation: E.S., T.B. and G.V.; Analysed NGS data: P.W.; Manuscript writing: E.S., T.B., A.P. and G.V.; Final approval of manuscript: E.S., P.W., P.S., M.M., A.A.P., F.H., S.G., G.R.R., Q.Z., A.F., T.B., J.Z., H.T., H.C., R.C.B., A.P. and G.V.; Accountable for all aspects of the study: G.V.

CONFLICT OF INTEREST STATEMENT

A.A.P., Honoraria from Sobi, AbbVie, BMS| Research support from Pfizer, Sumitomo, Kronos Bio; E.S., P.W., P.S., M.M., F.H., S.G., G.R.R., Q.Z., A.F., T.B., J.Z., H.T., H.C., R.C.B., A.P. and G.V. No disclosure(s).

PRIOR PRESENTATION

Portions of these data were presented at the annual meeting of the United States Canadian Academy of Pathology (USCAP) meeting in 2023 held in New Orleans, LA, USA, with some germline data presented as a moderated presentation at the 2024 USCAP meeting in Baltimore, MD, USA.

Supporting information

Data S1.

Symes EO, Wang P, Sojitra P, Menon MP, Patel AA, Hasan F, et al. Somatic co‐alteration signatures are prognostic in high‐grade TP53 ‐mutated myeloid neoplasms. Br J Haematol. 2025;206(4):1103–1108. 10.1111/bjh.19895

DATA AVAILABILITY STATEMENT

Please address data requests to the corresponding author via email: girish.venkataraman@uchospitals.edu.