CDKN2A Deletions Define an Unfavorable Subgroup within the MYD88/CD79B (MCD) Subtype of Diffuse Large B-Cell Lymphoma (DLBCL) and Are Mutually Exclusive with TP53 mutations

Background: Alterations (particularly biallelic deletions) of the tumor suppressor gene CDKN2A are frequent in the ultra-aggressive lymphoblastic (Quesnel et al, Blood 1995) and Burkitt lymphomas (Schmitz et al, Nature 2012). They also occur in DLBCL, and in prior studies they were associated with poor prognosis in conjunction with TP53 mutations (Jardin, Blood 2010). However, recent genomic classifications of DLBCL have noted frequent CDKN2A alterations in the MCD subtype (characterized by MYD88^L265P and CD79 mutations; Wright et al, Cancer Cell 2020-LymphGen classifier). MCD tumors show propensity for extranodal invasion, immune evasion, and are enriched among relapsed/refractory DLBCL (Ollila et al, Blood 2021). There is an interest in targeting the MCD subgroup with novel treatment approaches, but prognostic factors specific to MCD DLBCL are uncertain. We examined the association between CDKN2A deletions and other mutations, genomic subtypes, and prognosis in DLBCL.

Methods: We selected DLBCL cases submitted for next generation sequencing (NGS) as part of routine clinical care (FoundationOne Heme assay, Foundation Medicine, Inc., Cambridge, MA). All samples underwent central review by a board-certified pathologist. NGS was performed on hybridization-captured, adaptor ligation-based libraries in up to 405 cancer-related genes (Frampton et al, Nat Biotechnol, 2013), identifying clinically relevant base pair substitutions, indels, copy number alterations, and rearrangements. Co-occurrence/exclusivity was evaluated by odds ratios (OR) with P-values corrected for multiple testing using false discovery rate (FDR). Prognostic analysis was performed using publicly available data from the Haematological Malignancy Research Network (HMRN) study of 648 patients treated with RCHOP chemotherapy for DLBCL (Lacy et al, Blood 2020).

Results: Among 165 patients with confirmed DLBCL, median age was 67 (interquartile range, 56-76), and 48% were women. Biopsies were from an extranodal site in 113 cases (68%). CDKN2A alterations were present in 42 samples (25%): most commonly biallelic deletions (N=34), short variant alterations (N=7), and 1 rearrangement. CDKN2A deletions were found in 28 (25%) of extranodal and 6 (12%) of nodal biopsies (Fisher's exact P=.06). MYC-IGH rearrangement was detected in 3 (7%) of tumors with CDKN2A deletions and 5 (4%) of those without them (P=.42), but BCL2-IGH rearrangement was rare in tumors with CDKN2A deletions (2% vs. 33%, respectively; P<0.001).

Mutations in only 3 genes were statistically significantly associated with CDKN2A deletions: MYD88 (OR=12.6, P_corr=3.9 x 10 ^-6), CD79B (OR=20.4, P_corr =.00031) were highly co-occurring, whereas TP53 (OR=0.09, P_corr=.0072) was highly mutually exclusive (Fig. A/B). Among tumors with CDKN2A deletions, 56% had mutations in MYD88, 32% in CD79B, and 32% in PIM1, but only 6% in TP53. Conversely, in DLBCL without CDKN2A deletions, TP53 mutations were present in 41%, while <10% had mutations in MYD88, CD79B, or PIM1.

When studied using the LymphGen DLBCL classifier, CDKN2A deletions were present in 14 out of 16 MCD (88%), 2 out of 10 (20%) BN2, 18 out of 111 (16%) of unclassifiable tumors, and in no tumors classified as A53, EZB, or ST2 (Fig. C; P<.001 for MCD vs others). CDKN2A deletions were also specific to the hc-MCD subtype using our simplified hierarchical classifier developed for multi-gene NGS panels (Fig. D).

In the HMRN data, CDKN2A deletions were observed in 10% of cases, significantly more often (34%) in the MYD88 cluster (corresponding to LymphGen MCD) than in other clusters (6.3%; P<.001). Conversely, TP53 alterations were significantly less frequent in the MYD88 cluster (7% vs 21% in others, P=.004). CDKN2A deletions were associated with significantly worse progression-free and overall survival (Fig. E/F) within the MYD88 cluster (independently of the International Prognostic Index), but not in others.

Conclusions: CDKN2A deletions are specific to the MCD genomic subtype of DLBCL and indicate particularly poor prognosis within this class. Relative mutual exclusivity with TP53 mutations suggests that CDKN2A deletion may constitute an alternative, critical "hit" to a tumor suppressor gene in MCD DLBCL. Further research should examine the clinical relevance of CDKN2A deletions for refractoriness to standard therapy and its role in immune evasion that is characteristic of relapsed/refractory MCD DLBCL.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal