To the Editor:

TP73, a TP53 family member, is involved in the pathogenesis of adult T-cell leukemia/lymphoma (ATL) and is regulated by the intragenic super-enhancer [1, 2]. We have reported that TP73 structural variants (SVs) with deletion of exons 2–3, which were potentially associated with the formation of super-enhancer, were present in a fraction of ATL patients. Experimental deletion of exons 2–3 conferred a competitive advantage to ATL cells by promoting their proliferation [2]. Hence, following this observation, here we explored the clinical landscape of TP73 SVs in ATL patients.

Study subjects were mogamulizumab-naïve ATL patients without prior allogeneic hematopoietic stem cell transplantation, who then received mogamulizumab-containing treatment (n = 63) [3]. They had all been enrolled in our previous study [4]. DNA/RNA preparation for genomic analysis, exome library preparation and sequencing, RNA-sequencing, somatic variant call, fusion gene detection using RNA-sequencing, detection of SVs in the 3′-UTR of CD274 gene, HLA alteration call, and statistical analyses were all as previously described [4]. Tumor purity for tumor-derived DNA was calculated by Control-FREEC [5]. TP73 SVs, of exons 2 or 2–3 deletion, were evaluated by three different programs, such as DeviCNV v1.5.1 (Rhelixa Inc., Tokyo, Japan) [6], Manta v1.6.0, and GRIDSS2 v2.13.2, using our previous exome dataset [4]. The presence of TP73 SVs was defined when they were detected by at least one of these three programs. In the cases where TP73 SVs were identified by any of these programs, the original sequence data were manually investigated to confirm the presence of TP73 SVs. The current study was approved by the institutional review boards at all participating sites, and all patients provided written informed consent before blood or tissue sampling.

As a result, patient 1, 2, and 3 were determined to harbor exon 2 deletion, and patients 5, and 7 were determined to harbor exons 2–3 deletion, by DeviCNV (Supplementary Table S1). Although, in these five cases, TP73 SVs were not detected by Manta or GRIDSS2 (Supplementary Table S2 or S3, respectively), the presence of TP73 SVs was confirmed manually (Supplementary Fig. S1A, S1B, S1C, S1E, and S1G). The discrepancy in result is likely due to the difference in detection methods (e.g., exon-level detection vs breakpoint detection). 　Patient 4 was determined to harbor TP73 SVs by all three programs, which was confirmed manually (Supplementary Fig. S1D). Patient 6 was determined to harbor exons 2–3 deletion by DeviCNV, and the deletion included a part of exon 4 by both Manta and GRIDSS2. The manual investigation determined this case to harbor TP73 SVs (Supplementary Fig. S1F). Although patient 8 was not determined to harbor TP73 SVs by DeviCNV, it was determined by both Manta and GRIDSS2, which was manually confirmed (Supplementary Fig. S2). Patients 9 and 10 were determined to harbor only TP73 exon 3 deletion by DeviCNV. The manual investigation confirmed these both two to harbor only a part of exon3 deletion (patients 9 and 10, Supplementary Figs. S3 and S4, respectively), thus these were not included in the TP73 SVs (+) group. Collectively, TP73 SVs were found in 8 patients (13%), namely patients 1–8.

As for clinical characteristics, there were no significant differences between patients with or without TP73 SVs in terms of age, sex, previous systemic chemotherapy, Eastern Cooperative Oncology Group performance status or clinical subtype. On the other hand, serum LDH was significantly higher in patients with SVs than in those without (Supplementary Table S4). Notably, TP73 SVs held prognostic significance: median progression-free survival (PFS) of patients with TP73 SVs was 0.2 years versus 1.1 years for those without (P = 0.006; Fig. 1A, left panel). In addition, the median overall survival (OS) of the patients with or without TP73 SVs was 0.8 and 1.8 years, respectively (P = 0.004; Fig. 1A, right panel). Then, we performed multivariate analysis of PFS of the 63 ATL patients using six variables as follows: TP73 SVs, TP53, CCR4, and CD274 alterations, or clinical subtype and serum LDH. These three gene alterations were found to be significantly associated with PFS in our previous univariate analysis [4], and these two clinical variables were generally acceptable prognostic factors in ATL [7]. As a result, TP73 SVs were significantly associated with PFS (Fig. 1B). Multivariate analysis also for OS was performed using seven variables, such as TP73 SVs, and TP53, CCR4, CD28, and CD274 alterations, because the latter four gene alterations were significantly associated with OS previously in univariate analysis [4], or clinical subtype or serum LDH. Eventually, TP73 SVs were also significantly associated with OS (Fig. 1C). Collectively, these indicate that TP73 SVs are significantly and independently associated with a worse prognosis for ATL patients.

Fig. 1: Prognostic significance of TP73 SVs in patients with adult T-cell leukemia/lymphoma.
figure 1

A Progression-free survival (PFS) of adult T-cell leukemia/lymphoma (ATL) patients with or without TP73 SVs (left panel). Overall survival (OS) of ATL patients with or without TP73 SVs (right panel). B Multivariate analysis including TP73 SVs for PFS in patients with ATL. C Multivariate analysis including TP73 SVs for OS in patients with ATL.

Full size image

Next, we sought relationships between the presence of TP73 SVs and the 34 ATL driver gene alterations that were found in more than two patients each (Supplementary Table S5) [4]. This analysis revealed that patients with TP73 SVs significantly more frequently harbored TBL1XR1, or RHOA alterations, but there were no significant differences in the remaining 32 genes (Supplementary Table S6).

Using RNA-sequencing, we then analyzed the transcriptome profiles in ATL stratified according to TP73 SVs (cBioinformatics Inc., Tokyo, Japan). First, we confirmed that there were no significant differences in tumor purity between two groups (76.3% [median], and 57.2–99.1% [range] in eight patients with SVs, and 86.6%, and 52.9–100.0%, in 55 patients without SVs, P = 0.137). We previously reported that consensus clustering analysis of the present patients defines four transcriptome subtypes (TS) designated A, B, C, and D [4]. Here, we found no significant associations of TP73 SVs with a particular TS: A (1/19), B (3/18), C (0/5), and D (4/21) (P = 0.551). We then performed individual gene expression analysis, which revealed that TP73 expression was higher in patients with TP73 SVs than in those without (Fig. 2A, left panel). Similarly, the expression of TP73-AS3 (NCBI/Entrez Gene ID: 105378610), which is located on the anti-sense strand within the intragenic super-enhancer region of TP73, was higher in patients with SVs (Fig. 2A, right panel) (Supplementary Table S6). This finding is consistent with our previous report that TP73 SVs were associated with the activation status of TP73 gene transcription [2].

Fig. 2: Transcriptome profiles according to TP73 SVs.
figure 2

A TP73 expression is higher in patients with TP73 SVs (dark gray circles, right side, n = 8) than in those without (light gray circles, left side, n = 55) (median, 260.1 vs. 13.8 TMM [trimmed mean of M values], P = 0.001 [Mann–Whitney U test]) (upper left panel). TP73-AS3 expression is higher in patients with TP73 SVs than in those without (median, 6.8 vs. 0 TMM, P = 0.011) (upper right panel). B RAB26 expression is higher in patients with TP73 SVs (dark gray circles, right side) than in those without (light gray circles, left side) (median, 15.5 vs. 2.1 TMM, respectively, P < 0.001) (upper left panel). FER1L4 expression is also higher in patients with TP73 SVs (median, 11.8 vs. 2.8 TMM, respectively P = 0.034) (upper middle panel). DAB2 expression is lower in patients with TP73 SVs compared to those without (median, 7.8 vs. 45.8 TMM, P = 0.013) (upper right panel), as is ARL11 expression (2.3 vs. 13.3 TMM, P = 0.045) (lower left panel), PTK6 expression (1.8 vs. 10.3 TMM, P = 0.025) (lower middle panel), and TCAE3 expression (1.3 vs. 12.6 TMM, P = 0.044) (lower right panel). The horizontal straight and dotted lines represent the median value of gene expression (TMM, trimmed mean of M-values) in patients with and without TP73 SVs, respectively. C ABLIM1 expression is lower in patients with TP73 SV than in those without (median, 170.1 vs. 451.1 TMM, P = 0.014) (left panel). Also, LZTS2 expression is lower in patients with TP73 SVs compared to those without (median 87.1 vs. 149.3 TMM, P = 0.001).

Full size image

Next, of the 228,048 transcripts in GENCODE Release 34, we selected 28 genes which were upregulated or 60 downregulated in ATL cells with TP73 SVs, compared to those without SVs, based on our pre-determined criteria (Supplementary Table S8). From this selection, six notable findings emerged, namely, 1) RAB26 expression was higher in patients with TP73 SVs than in those without (Fig. 2B, upper left panel). This is consistent with the fact that RAB26 contributes to the progression of non-small cell lung cancer [8]. 2) A long non-coding RNA, FER1L4, expression was higher in patients with SVs (Fig. 2B, upper middle panel), consistent with the fact that FER1L4 is the direct transcriptional target of TP73 [9]. In contrast, expressions of 3) DAB2 [10], 4) ARL11 [11], 5) PTK6 [12], and 6) TCAE3 [13], were lower in patients with SVs (Fig. 2B, upper right, lower left, lower middle, and lower right panels, respectively). These four genes have been reported to be tumor suppressors, and especially DAB2 is associated with c-Myc downregulation [10].

Other than these selected genes, we also paid special attention to those genes that were regulated by experimental deletion of TP73 exons 2–3 in an ATL cell line (TL-Om1) in our previous study [2]. Among those genes, notably ABLIM1 and LZTS2 expression was lower in primary ATL cells with SVs than without (Fig. 2C, left and right panels, respectively, and Supplementary Table S6). It has been reported that overexpression of ABLIM1 in glioblastoma cells leads to an attenuated proliferation [14], and deletion of LZTS2 increases susceptibility to spontaneous and carcinogen-induced tumor development [15], thus suggesting that downregulation of these genes contributes to the aggressiveness of ATL. Collectively, these gene expression profiles differing between patients with or without TP73 SVs are consistent with our previous findings in which ATL cells with TP73 SVs acquired enhanced resistance to apoptosis and a growth advantage [2]. In addition, these gene profiles are also consistent with the present miserable prognosis of ATL patients with TP73 SVs.

To the best of our knowledge, this is the first report to describe the clinical landscape of TP73 SVs in patients with ATL, or indeed any type of cancer. Although the present investigation offers significant observations regarding TP73 SVs, an unavoidable limitation should be recognized. The presence or absence of TP73 SVs was evaluated using three different programs. Because varying tumor purity of the samples cannot be integrated into this program, despite our confirmation of the lack of significant differences in this respect, it cannot be completely excluded that the occurrence of TP73 SVs may have been underestimated in the present study.

In conclusion, the presence of TP73 SVs was significantly and independently associated with a worse prognosis of ATL patients receiving mogamulizumab-containing treatment. ATL cells with TP73 SVs exhibit gene expression profiles associated with enhanced resistance to apoptosis and growth advantage. Taken together, we believe that the current results contribute to a better understanding of the pathogenesis not only of ATL, but also many types of cancer associated with TP73, and particularly with TP73 SVs.