Impact of TP53 Genomic Alterations in Large B-Cell Lymphoma Treated With CD19-Chimeric Antigen Receptor T-Cell Therapy

PURPOSE

Tumor-intrinsic features may render large B-cell lymphoma (LBCL) insensitive to CD19-directed chimeric antigen receptor T cells (CAR-T). We hypothesized that TP53 genomic alterations are detrimental to response outcomes in LBCL treated with CD19-CAR-T.

MATERIALS AND METHODS

Patients with LBCL treated with CD19-CAR-T were included. Targeted next-generation sequencing was performed on pre–CAR-T tumor samples in a subset of patients. Response and survival rates by histologic, cytogenetic, and molecular features were assessed. Within a cohort of newly diagnosed LBCL with genomic and transcriptomic profiling, we studied interactions between cellular pathways and TP53 status.

RESULTS

We included 153 adults with relapsed or refractory LBCL treated with CD19-CAR-T (axicabtagene ciloleucel [50%], tisagenlecleucel [32%], and lisocabtagene maraleucel [18%]). Outcomes echoed pivotal trials: complete response (CR) rate 54%, median overall survival (OS) 21.1 months (95% CI, 14.8 to not reached), and progression-free survival 6 months (3.4 to 9.7). Histologic and cytogenetic LBCL features were not predictive of CR. In a subset of 82 patients with next-generation sequencing profiling, CR and OS rates were comparable with the unsequenced cohort. TP53 alterations (mutations and/or copy number alterations) were common (37%) and associated with inferior CR and OS rates in univariable and multivariable regression models; the 1-year OS in TP53-altered LBCL was 44% (95% CI, 29 to 67) versus 76% (65 to 89) in wild-type (P = .012). Transcriptomic profiling from a separate cohort of patients with newly diagnosed lymphoma (n = 562) demonstrated that TP53 alterations are associated with dysregulation of pathways related to CAR-T-cell cytotoxicity, including interferon and death receptor signaling pathway and reduced CD8 T-cell tumor infiltration.

CONCLUSION

TP53 is a potent tumor-intrinsic biomarker that can inform risk stratification and clinical trial design in patients with LBCL treated with CD19-CAR-T. The role of TP53 should be further validated in independent cohorts.

INTRODUCTION

Although most patients with newly diagnosed diffuse large B-cell lymphoma (DLBCL) are cured with frontline chemoimmunotherapy,^1,2 approximately 30% have relapsed or refractory (R or R) disease. Historical outcomes in R or R DLBCL have been poor.³ Autologous CD19-directed chimeric antigen receptor T-cell therapy (CAR-T) has resulted in unprecedented response rates of more than 70% in this population. However, more than 60% of CD19-CAR-T recipients will ultimately progress.^4-8

CONTEXT

• Key Objective

• Tumor-intrinsic biomarkers guiding prognostic assessment of large B-cell lymphoma (LBCL) treated with CD19-chimeric antigen receptor T-cell (CAR-T) therapy remain ill-defined. This observational study examined the prognostic role of TP53 alterations, identified by DNA sequencing, in LBCL treated with CD19-CAR-T. To our knowledge, this is the first study evaluating TP53 in this context.

• Knowledge Generated

• TP53 alterations (mutations and/or copy number alterations) emerge as potent determinants of response and overall survival after CD19-CAR-T therapy. Transcriptomic profiling in a separate cohort of patients with LBCL suggests that TP53 alterations distort tumor cellular mechanisms used by CAR-T cells to exert cytotoxicity.

• Relevance (J.W. Friedberg)

• TP53 genomic status in LBCL identifies patients potentially at increased risk for CD19-CAR-T therapy failure and, if these results are validated, should inform the design of future clinical trials.*

• *Relevance section written by JCO Editor-In-Chief Jonathan W. Friedberg, MD.

Tumor-extrinsic factors contributing to CAR-T failure include T-cell exhaustion and persistence, circulating monocytic myeloid-derived suppressor cells, and an immunosuppressive tumor microenvironment (TME).^9-13 By contrast, tumor-intrinsic factors are not well characterized. CD19 antigen evasion accounts for roughly 30% of large B-cell lymphoma (LBCL) treated with axicabtagene ciloleucel.¹⁴ As an acquired therapy–induced event, antigen evasion offers limited opportunity for upfront risk stratification strategies.¹⁵ Recent studies suggest that tumor cells may become insensitive to CAR-T cells via intrinsic resistance mechanisms, rendering them insensitive to apoptosis and cytokine-triggered killing.^10,16-18 How the pre-existing genomic landscape of LBCL contributes to these intrinsic resistance mechanisms is unclear.

Traditional tumor-related features carrying prognostic importance in newly diagnosed LBCL (eg, activated B-cell–like, cell of origin [COO] phenotype, and double- or triple-hit translocations) are not informative in R or R disease treated with CD19-CAR-T.^5,7,19 Therefore, readily available tumor biomarkers that can predict response, guide patient selection, and inform resistance mechanisms are needed. Alterations in the p53 gene (TP53) are an established marker of poor outcome across malignancies,^20-22 including LBCL.^23,24 TP53's role in CD19-CAR-T recipients is unknown. P53 regulates apoptosis and may also play a role in immune evasion and induction of an immunosuppressive TME,^25,26 all of which are mechanisms affecting CAR-T-cell cytotoxicity.²⁷ We hypothesize that genomic alterations in TP53 contribute to failure of CAR-T therapy in LBCL. In this observational study, we explored disease-related determinants of CD19-CAR-T success in LBCL. Genomic alterations in TP53, including mutations and copy number alterations (CNAs), emerged as potent predictors of response and overall survival (OS). Transcriptomic analyses elucidated mechanisms by which p53 may modulate CAR-T efficacy.

MATERIALS AND METHODS

Study Population

This retrospective analysis included adult patients with primary or transformed LBCL treated at Memorial Sloan Kettering Cancer Center (MSKCC) with autologous CD19-CAR-T cells (axicabtagene ciloleucel [axi-cel], tisagenlecleucel [tisa-cel], and lisocabtagene maraleucel [liso-cel]) between April 28, 2016, and April 21, 2021. Axi-cel and tisa-cel were given as standard therapy and liso-cel under the TRANSCEND NHL 001 study (NCT02631044).⁵ Patient data were entered and stored in a REDCap database.²⁸ Research was approved by the Institutional Review Board committee and conducted in accordance with the Declaration of Helsinki. Informed consent was obtained for tumor sequencing.

Definitions

Tumor biopsies were reviewed at MSKCC. Large-cell lymphoma and double-expressor classification followed standard criteria.²⁹ COO was determined by the Hans algorithm.³⁰

All patients received CAR-T cells after lymphodepleting chemotherapy following the US Food and Drug Administration label.^5-7 Response assessment was performed according to the Lugano criteria.³¹ Post–CAR-T response was defined relative to the last disease assessment before the infusion. Best response reflects the best response achieved up to 90 days after cell infusion. Progression-free survival (PFS) was defined as the time from CAR-T infusion to first documented progression or death from any cause. The Data Supplement (online only) lists further definitions.

Sequencing

HemePACT-targeted sequencing was performed as previously described^22,32 on tumor samples from 82 patients. Briefly, DNA was extracted from formalin-fixed paraffin-embedded tumor tissue and patient-matched noninvolved samples (available in 73 of 82 patients). Bar-coded libraries were generated, and sequencing was performed for all exons and select introns across a custom gene panel (Data Supplement) of 401 genes (n = 76, v.1) or 412 genes (n = 6, v.2). Sequences were analyzed via a custom pipeline to identify somatic alterations, including mutations, fusions, and CNAs.^22,32-34

Transcriptome Profiling of TP53-Altered DLBCL

RNA sequencing, mutation annotations, and CNAs were obtained from the GDC Data Portal (dbGAP # phs001444, phs001175, and phs000178; Data Supplement).

To investigate the relationship between TP53 alterations and the interferon (IFN) pathway, we calculated an IFN score by averaging the normalized gene expression of the Hallmark set of IFN-γ response genes. We then compared the distribution of altered genes (mutations and/or homozygous deletions) between IFN-high (upper tercile) and IFN-low (lower tercile) patients using Fisher's exact test with Benjamini-Hochberg correction for multiple comparisons.

Three methods of immune cell deconvolution were used (Data Supplement) to characterize the TME by TP53 status.^35-37

Statistical Analysis

Descriptive statistics, including median and interquartile range (IQR) for continuous variables and percentages for categorical variables, are provided. Fisher's exact test or χ² test was used to evaluate the association between categorical variables. The Wilcoxon rank-sum test or Kruskal-Wallis test was used to assess the difference in a continuous variable between or among patient groups. Univariable and multivariable logistic regression and Cox regression models were constructed to evaluate associations with outcomes (Data Supplement).

RESULTS

Patient Characteristics

A total of 153 patients (Table 1; Fig 1A) with R or R LBCL treated with CD19-CAR-T (axi-cel 77 [50%], tisa-cel 49 [32%], and liso-cel 27 [18%]) were included. At the time of infusion, the median patient age was 66 years (IQR 57-72). Twenty-eight patients (18%) had high-grade lymphoma histology, with double- or triple-hit cytogenetic translocations identified in 22 patients. Transformed lymphoma, nongerminal center B-cell COO, and double expressor immunophenotype were common. Before apheresis, patients were heavily pretreated (> 3 lines; 84, 55%); 50 (33%) had primary refractory disease. Bulky disease (> 10 cm) and disease stage III-IV at apheresis were observed in 17 (11%) and 106 (70%) patients, respectively.

TABLE 1.

Population Characteristics

FIG 1.

Outcomes after CD19-CAR-T in LBCL. (A) Flow diagram describing the population (cohort structure). A total of 153 patients were included; 82 patients had undergone tumor sequencing (HemePACT) before CAR-T infusion. (B) Alluvial plot representing response to CAR-T over time. Percent response is calculated from patients with evaluable disease at time of assessment. (C-E) Cumulative incidence of loss of response for responding patients (CR or PR), OS, and PFS from the day of CAR-T infusion. (F) Landmark analysis, from 28 days after CAR-T infusion, for OS by best response to CAR-T. CAR-T, chimeric antigen receptor T-cell; CNA, copy number alterations; CR, complete response; LBCL, large B-cell lymphoma; LoR, loss of response; NE, nonevaluable; OS, overall survival; PD, progressive disease; PFS, progression-free survival; PR, partial response; R or R, relapsed or refractory; SD, stable disease.

The median follow-up time after CAR-T infusion was 21.1 months (IQR 10.7-30.7 months). Among patients with disease response evaluated on day 28 (±7 days), 62 (46%) and 47 (35%) attained a complete response (CR) and partial response (PR). As of 90 days after infusion, the overall best response rate was 79% (CR 54%, PR 25%; Fig 1B). Disease response did not improve by day 90 in patients with stable disease or progressive disease (SD or PD) on day 28 (±7). However, of 47 patients with PR by day 28, 12 (26%) transitioned to CR by day 90. At 1 year, the cumulative incidence of response loss (Fig 1C) was 40% (95% CI, 28 to 51) in patients achieving a CR versus 68% (50 to 81) with a PR (Gray's P value < .001).

The median OS and PFS from CAR-T infusion were 21.1 months (95% CI, 14.8 to not reached) and 6 months (3.4 to 9.7; Figs 1D and 1E). The corresponding 1-year OS and PFS were 65% (57 to 73) and 38% (30 to 47). At 1 year, the cumulative incidence of relapse or progression was 56% (48 to 64). In a day-28 landmark analysis, OS was longest in patients attaining CR (Fig 1F).

Overall, our population is heavily pretreated and results mirror CAR-T registration trials.^5-7

Spectrum of TP53 Alterations

Targeted next-generation sequencing using MSK-HemePACT, at a median coverage of ×762 (IQR 664-851), was performed in 82 patients (54%; Fig 1A). Samples were collected at a median of 141 days (IQR 78–315 days) before CAR-T infusion. Cohorts with and without sequencing were comparable in terms of patient characteristics (Table 1) as well as CR rates (χ² P value = .9775) and OS (Log-rank P value = .831).

Alterations in TP53 (37%), KMT2D (MLL2; 44%), BCL2 (41%), and CDKN2A (37%) were common. Using the LymphGen classifier for determining the DLBCL genetic subtype,³⁸ 44% of tumors were classified as other, followed by EZB (42%). The remaining subtypes accounted for < 6% of patients each.

Twenty-five patients had a total of 28 mutations in the TP53 gene (23 missense, two in-frame deletions, two splice sites, and one truncating). Five patients had CNAs without mutations (Fig 1A; Data Supplement). All TP53-CNAs were classified as homozygous deletions. TP53-altered and wild-type (WT) LBCL generally had similar clinical features (Data Supplement), although prelymphodepletion lactate dehydrogenase (LDH) was higher in TP53-altered LBCL (P = .002). TP53 mutations were almost exclusively located in the DNA binding domain (27 of 28; Data Supplement). All mutations were categorized as oncogenic or likely oncogenic by OncoKB³⁹ except H214Q (1 of 28), classified as inconclusive.

Predictors of Outcomes

To determine whether features known before CAR-T infusion predict treatment success, we investigated associations between disease-related features and CAR-T outcomes. In a univariable analysis (Fig 3A), only TP53 alterations (odds ratio [OR], 3.48; 95% CI, 1.36 to 9.36; P = .011), primary refractory disease (OR, 2.31; 95% CI, 1.15 to 4.71; P = .019), and SD or PD at infusion (OR, 2.28; 95% CI, 1.07 to 5.11; P = .037) were associated with a lower likelihood of achieving CR. CR rates by day 90 were 33 of 51 (65%) versus 10 of 29 (34%), P = .009 in favor of TP53 WT (Fig 3B). TP53 alterations were also significantly associated with response in an ordinal logistic regression model considering a 3-level outcome (CR > PR > SD or PD; Data Supplement). TP53 alterations remained an independent predictor of response (OR, 3.61; 95% CI, 1.31 to 10.7; P = .016) in a multivariable model adjusting for age, Karnofsky performance status, primary refractory disease, LDH, and CAR-T costimulatory domain (Table 2). TP53 alterations were not correlated with grade ≥ 2 cytokine release syndrome (P = .76) or grade ≥ 2 immune effector cell–associated neurotoxicity syndrome (P = .11).

FIG 3.

TP53 as a determinant of response and survival after CD19-CAR-T in LBCL: (A) univariable logistic regression and Cox regression of disease-related feature for CR and OS, respectively; (B) best CR rates up to day 90 after CAR-T infusion by TP53 status; (C) OS by TP53 status; and (D) OS by TP53 status and CAR-T costimulatory domain (CD28: axicabtagene ciloleucel; 41BB: tisagenlecleucel and lisocabtagene maraleucel). CAR-T, chimeric antigen receptor T-cell; CR, complete response; HR, hazard ratio; LBCL, large-B-cell lymphoma; LDH, lactate dehydrogenase; NHL, non-Hodgkin lymphoma; OR, odds ratio; OS, overall survival; WT, wild-type.

TABLE 2.

Multivariable Analysis of Response and OS

In a univariable Cox regression model for OS (Fig 3A), the risk of death was greater in TP53-altered versus WT LBCL (hazard ratio [HR], 2.19; 95% CI, 1.18 to 4.10). MYC rearrangement, high proliferative index (KI-67 > 75%), bulky disease, elevated LDH, and nonresponding disease at infusion were also associated with poorer survival. The 1-year survival rate with TP53-altered versus WT lymphoma was 44% (29 to 67) and 76% (65 to 89), respectively (P = .012; Fig 3C). In a multivariable Cox regression model, TP53 remained independently associated with inferior survival (HR, 2.03; 95% CI, 1.02 to 4.03; P = .044).

TP53 alterations were not associated with PFS in univariable (HR, 1.40; 95% CI, 0.80 to 2.42; P = .24) or multivariable (HR, 1.51; 95% CI, 0.81 to 2.81; P = .19) Cox regression models (Data Supplement).

TP53 and CDKN2A belong to the TP53 pathway. Notably, genomic alterations in these genes were mutually exclusive (Fig 2 P = .004). In contrast to TP53, alterations in CDKN2A were primarily CNAs and were not associated with lower CR and OS rates (Data Supplement).

FIG 2.

Oncoprint representing genomic landscape of LBCL among CD19-CAR-T recipients. Top 10% aberrated genes were included. TP53 and CDKN2A are mutually exclusive (P = .002). CAR-T, chimeric antigen receptor T-cell; CR, complete response; FL, follicular lymphoma; GCB, germinal center B cell; LBCL, large B-cell lymphoma; NE, not evaluated; NHL, non-Hodgkin lymphoma.

Collectively, our findings suggest that TP53 status in LBCL is a prognostic marker among CD19-CAR-T recipients.

TP53 and CAR-T Product

The choice of CAR-T product is a modifiable treatment factor. Therefore, we performed an exploratory analysis on the impact of TP53 molecular status by the CAR-T product categorized by the costimulatory domain. In a multivariable logistic regression adjusting for factors contributing to product selection, including age, Karnofsky performance status, and primary refractory disease, the likelihood of attaining CR by day 90 in patients with TP53-altered lymphoma was independent of the CAR-T costimulatory domain (Data Supplement). However, treatment of TP53-altered LBCL with a 41BB product was associated with lower OS and PFS compared with CD28 (ie, axi-cel). In a multivariable Cox regression model (Data Supplement), these corresponded to hazard ratio is of 3.82 (1.31 to 11.1) and 4.57 (1.64 to 12.8). The OS and PFS at 1 year were 36% (17 to 77) and 10% (1.7 to 62) with 41BB CAR-T versus 51% (32 to 83) and 34% (17 to 69) with axi-cel (Fig 3D). Although preliminary and requiring further validation, our findings suggest that TP53 status may have implications for choosing the CAR-T product.

TP53 Alterations are Associated With Disrupted Apoptosis and IFN Signaling

On the basis of the current understanding of CAR-T killing mechanisms,^27,40 we hypothesized that TP53 alterations might impair LBCL susceptibility to CAR-T by the following mechanisms: (1) impairment of IFN signaling,^10,41,42 (2) dysregulation of the extrinsic apoptotic pathway,^16,18,40 (3) imposition of an immunosuppressive TME,^25,26 and (4) downregulation of CD19.⁴³ We obtained tumor RNA sequences of 562 patients with newly diagnosed DLBCL to evaluate these four hypotheses. Patients with altered TP53 (mutations and/or homozygous deletions, n = 148 of 562 [26%]) exhibited significant downregulation of the IFN (α and γ) and apoptosis pathways compared with TP53 WT patients (n = 414 of 562 [74%], Fig 4A, Data Supplement). Specific genes that were downregulated in TP53-altered DLBCL suggest that both granzyme-perforin and cytokine-mediated CAR-T killing mechanisms could be impaired, alongside impairment of the extrinsic apoptotic pathway (Fig 4B). When comparing low-IFN-γ and high-IFN-γ gene expression signatures, alterations in TP53 were more likely in the low group (OR = 2.6, P < .0001, Fig 4C, Data Supplement). Notably, TP53 was the only gene significantly differentiating between groups. This suggests that TP53 plays a functional role in IFN signaling. We did not find a correlation between CD19 expression and TP53 alteration status (Data Supplement). Finally, inference of TME composition from RNA-seq data by three different methods (Data Supplement) demonstrated upregulation of CD8+ T cells and a subpopulation of memory CD8+ T cells in TP53 WT DLBCL patients (Fig 4D). Thus, TP53 alterations may induce an immunosuppressive TME, impairing CAR-T-cell ability to infiltrate tumor sites. In summary, inhibition of the IFN response, apoptosis, and reduced CD8+ T-cell infiltration are all potential mechanisms contributing to inferior outcomes in patients with TP53-altered LBCL treated with CAR-T.

FIG 4.

Transcriptomic profiling of TP53-altered DLBCL. Transcriptomic profiling was performed on 562 tumor samples from patients with newly diagnosed DLBCL. (A) GSEA of genes ranked by their t-statistic in comparison between TP53-altered (n = 148) and WT (n = 414) patients demonstrated under-representation of IFN-γ response and apoptosis genes (Hallmark) in TP53-altered DLBCL. (B) Heatmap of differentially expressed genes (FDR <0.01) between TP53-altered and WT DLBCL. Hallmark IFN-γ response and apoptosis genes were studied. (C) IFN-γ signature was categorized into terciles on the basis of normalized gene expression. TP53 alterations were over-represented in tumors with low IFN-γ signature compared with the highest tercile. (D) Immune cell deconvolution was performed using three different deconvolution tools (Data Supplement). These tools infer the cellular components within a tumor on the basis of their RNA profile. All methods demonstrated reduced CD8 T-cell infiltration in TP53-altered tumors. All immune cell fractions were compared using a nonparametric Wilcoxon test with Benjamini-Hochberg correction and a Q-value threshold of 0.05 for statistical significance. CTSP, Clinical Trials Sequencing Project; DLBCL, diagnosed diffuse large B-cell lymphoma; FDR, false discovery rate; GSEA, Gene Set Enrichment Analysis; IFN, interferon; NCICCR, The National Cancer Institute's Center for Cancer Research; TCGA, The Cancer Genome Atlas; WT, wild-type.

DISCUSSION

We address a critical clinical question: are there readily available histologic and molecular features that predict CD19-CAR-T success or resistance. To the best of our knowledge, we are the first to describe TP53 genomic alteration as a potent predictor of CAR-T failure. To investigate mechanisms potentially driving inferior response and survival in TP53-altered lymphoma treated with CAR-T, we explored gene expression patterns in LBCL. Our findings suggest that TP53 alterations affect cellular processes, including downregulation of IFN, inhibition of the extrinsic apoptosis pathway, and induction of an immunosuppressive TME. Importantly, CAR-T cytotoxicity is mediated by all these mechanisms. Taken together, TP53 emerges as a tumor-intrinsic factor informing both prognosis and potential mechanisms of resistance in R or R LBCL treated with CD19-CAR-T.

Histologic and molecular biomarkers in LBCL have been instrumental in risk profiling, understanding resistance mechanisms, and more recently, treatment selection.^44,45 Traditional prognostic markers in patients with newly diagnosed LBCL have failed to maintain their clinical significance in our cohort and other studies.^5,7,19 Therefore, identification of TP53 as a determinant of response and survival fills a clinical gap. TP53 alterations were not well-linked to disease phenotype, aside from elevated LDH, which lacks specificity. Thus, sequencing should be pursued in most patients. Importantly, TP53 alterations may be actionable. Strategies targeting attenuated p53, such as restoring WT p53 function, show promise in LBCL.⁴⁶ Combining these strategies with CAR-T could potentially sensitize TP53-altered tumors to treatment. Intriguingly, we observed that OS and PFS in TP53-altered LBCL were higher with axi-cel compared with 41BB products. These findings raise the question whether cytotoxic pathways induced by the CD28 costimulatory domain can override mechanisms of TP53-mediated tumor evasion.^47,48 Importantly, further validation in larger cohorts and prospective trials is warranted to guide CAR-T product selection. Altogether, given the prognostic (biomarker informs outcome) and potentially predictive (biomarker informs therapeutic intervention) role of TP53 alterations, we argue that sequencing should be pursued for all CD19-CAR-T candidates.

Next-generation sequencing techniques have opened new avenues for LBCL genomic classification and identification of targetable pathways.^38,49-51 Genomic assignments were developed on newly diagnosed patients, which could explain the high rates of unassigned genomic subtypes in R or R LBCL, as demonstrated by us and others.⁵² Using the HemePACT panel, we observed that TP53 alterations are common (37%). In the newly diagnosed setting, rates are lower (approximately 20%),^23,52 indicating that TP53 alterations are enriched in resistant disease. TP53 and CDKN2A are components of the p53 pathway. Nevertheless, genomic alterations in either gene were mutually exclusive, suggesting that they are independent drivers of LBCL. Monti et al⁵³ reported that CNAs in genes involved in the p53 pathway are associated with poor outcomes in DLBCL. In our cohort, only a minority of patients had CNAs in TP53. CDKN2A aberrations, which were primarily CNAs, lacked prognostic significance. Collectively, our findings suggest that TP53 alterations independently influence outcomes in R or R LBCL.

In recent years, TP53 has emerged as a mediator of tumor immune escape.^25,26 Using genomic and transcriptomic profiling of newly diagnosed DLBCL tumor samples, we identified two complementary mechanisms, downregulation of IFN (α and γ) signaling and decreased CD8 tumor infiltration, which may contribute immunologic resistance to CAR-T in TP53-altered lymphoma. IFN plays an instrumental role in cytotoxic T-cell expansion, effector activity, and viability.^13,54 Loss of IFN signaling is associated with resistance to immune checkpoint blockade in melanoma.^42,55,56 Therefore, downregulation of IFN signaling could induce an immune-privileged TME, preventing the migration and activity of CAR-T cells. Importantly, upregulation of IFN can have opposing effects, potentiating antitumor response on the one hand and suppressing it on the other hand.^41,54 Jain and colleagues recently reported that higher expression of IFN signature genes (ISG.RS)⁴¹ is associated with lack of durable response in R or R LBCL treated with axi-cel.¹⁰ They elegantly show that high ISG.RS signaling is associated with higher expression of multiple T-cell inhibitory ligands. Differences in findings from our analysis could be attributed to the different gene sets studied. They also find upregulation of IFN-γ to be associated with favorable response, supporting the association that we observed between TP53 alterations and enrichment of the IFN-γ pathways.

Resistance to tumor cell–induced apoptosis is another potential mechanism underlying inferior outcomes with CAR-T in TP53-altered LBCL.^10,16,18,40 We identified downregulation of FAS-ligand, an inducer of extrinsic apoptosis via death receptors, in TP53-altered DLBCL. Upadhyay et al¹⁸ recently demonstrated that FAS and FAS-ligand interaction mediates antigen-dependent and antigen-independent CAR-T killing. Furthermore, two groups have reported loss of death receptor signaling as antigen-independent resistance mechanisms in B-lymphoblastic leukemia.^16,17 Therefore, impairment of tumor cell apoptosis could contribute to reduced efficacy of CAR-T cells. Priming tumor cells for apoptosis and stabilization of p53 may increase TP53-altered susceptibility to CAR-T.

This study has several limitations. First, sequencing of tumor samples was driven by clinical decision making, potentially leading to selection bias. However, we note that population features and outcomes were similar between sequenced and nonsequenced patients. Second, publicly available gene expression data coupled with TP53 alteration status were available only in newly diagnosed DLBCL. Treatment exposures over time could remodel the TME and alter T-cell phenotype and activity. Therefore, the relationship between genetic and transcriptional programs may vary in R or R LBCL. Finally, to our knowledge, although this is the first and largest analysis evaluating the role of TP53 in LBCL treated with CAR-T, sample size remains limited. External validation in a larger cohort is warranted. Furthermore, experimental in vitro and in vivo systems are required to establish a causal link between p53 dysregulation and CAR-T efficacy.

In conclusion, we show that TP53 aberrations are common in R or R LBCL and proved to be valuable prognostic markers in CD19-CAR-T recipients. Gene expression profiling of LBCL, coupled with our current understanding of CAR-T cytotoxicity, suggested that p53 dysfunction impairs mechanisms by which CAR-T cells exert their effector activity. Overall, our findings should be considered in evaluating the probability of CAR-T treatment success and clinical trial design targeting high-risk patients with TP53 alterations.

AUTHOR CONTRIBUTIONS

Conception and design: Roni Shouval, Ana Alarcon Tomas, Marcel van den Brink, Miguel-Angel Perales, Maria Lia Palomba

Financial support: Miguel-Angel Perales

Provision of study materials or patients: Aishat Olaide Afuye, Connie Lee Batlevi, Parastoo B. Dahi, Sergio A. Giralt, Richard J. Lin, Craig S. Sauter, Michael Scordo, Gunjan L. Shah, Miguel-Angel Perales, Maria Lia Palomba

Collection and assembly of data: Roni Shouval, Ana Alarcon Tomas, Aishat Olaide Afuye, Anna Alperovich, Theodora Anagnostou, Parastoo B. Dahi, Warren B. Fingrut, Richard J. Lin, Nishi Shah, Marcel van den Brink, Maria Lia Palomba

Data analysis and interpretation: Roni Shouval, Ana Alarcon Tomas, Joshua A. Fein, Jessica Flynn, Ettai Markovits, Shimrit Mayer, Michal J. Besser, Connie Lee Batlevi, Sean M. Devlin, Sergio A. Giralt, Gal Markel, Gilles Salles, Craig S. Sauter, Michael Scordo, Gunjan L. Shah, Ruth Scherz-Shouval, Marcel van den Brink, Miguel-Angel Perales, Maria Lia Palomba

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Impact of TP53 Genomic Alterations in Large B-Cell Lymphoma Treated With CD19-Chimeric Antigen Receptor T-Cell Therapy

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).