Novel immune checkpoint inhibitors (ICIs), including anti-cytotoxic T lymphocyte antigen 4 (CTLA-4), programmed death 1 (PD-1), and their ligands, PDL1 (CD274) and PD-L2 (CD273), have great potential for therapeutic efficacy in a variety of tumors, such as metastatic melanoma and non-small cell lung carcinoma (NSCLC), along with bladder urothelial cancer (1). However, patients treated with ICIs have variable response rates, and in several cases, a low treatment response rate has limited their practical application (2, 3). Therefore, current research is focused on how to predict patients who are responders and non-responders to ICI treatment (4). With our continual improvement in understanding genomic and microenvironmental processes associated with responses to ICIs, it becomes more beneficial for us to find biomarkers that could predict treatment effectiveness to ICIs, and it had also advantage in terms of supporting selection of patients and decision making by separating responders and non-responders accordingly (5, 6).

Poly (ADP-ribose) polymerases-1 (PARP1), a member of the PARP family, was identified as a DNA damage sensor that plays a critical role in the process of DNA damage repair, along with regulation of the expression of genes by poly(ADP-ribosyl), which is a sequence-specific DNA-binding transcription factor (7). Mutations in PARP1 have been found in several tumors, and some point mutations are closely linked to PARP1 inhibitor resistance (8). A recent study reported that PARP1 could ADP-ribosylate the regulatory T-cell (Treg)-specific transcription factor FOXP3, that negatively regulates the function of Treg cells, and PARP1 silencing can enhance PDL1 expression in tumor cells (9, 10). In preclinical experiments on breast cancer cell lines, small-molecule PARP1/PD-L1 inhibitor conjugates demonstrated more significant apoptosis and cytotoxic efficacy compared to their separate agents (11). These results suggest that not only could PARP1 alteration incorporate ICI treatment but also seem to potentially predict ICI therapeutic values. However, a systematic analysis of the prevalence and predictive value of PARP1 alteration for ICI therapeutic effectiveness in multiple tumor types remains largely unknown, and the associations between its expression levels and immunotherapy-associated signatures also need to be explored.

In our pan-cancer analysis study, we comprehensively investigated the frequency of PARP1 alterations and their predictive value in a variety of tumor types through an online database. There has been a frequency (2.9%) in PARP1 alterations and a significant predictive value for ICI treatments across more than 40,000 patients with various tumors. We analyzed the relationship between PARP1 alterations and tumor mutation burden (TMB) level, immune cell infiltrations, along with microsatellite instability (MSI) event. PARP1 alterations could predict the effectiveness of ICI therapies, as evidenced by our analysis results. Meanwhile, we performed a relationship analysis of PARP1 expression with TMB, MSI, and tumor immune infiltrations as well, based on the online database.

A variety of cancers are driven by genetic alteration, and their genomes contained an average of 4 to 5 driver mutations in combination coding with non-coding genomic elements (20). Another research showed that DNA repair defects could lead to some types of somatic mutation like BRCA1 or 2 (21). However, the role of PARP1 alteration across cancers remained unclear. In our study, we found PARP1 alterations as a variety of nonsynonymous mutations including missense, frameshift, splice site, nonsense, fusions, inframe and deletions, and total alteration rate of PARP1 gene was nearly 3% in 45604 patients with different tumors from TCGA database. Among all these patients, those with skin cancer, non-melanoma had the highest mutation frequency (8.98%, 44/490), followed by patients with endometrial carcinoma (5.29%, 31/586). In addition, our data showed that early-stage tumors were more easily detected by PARP1 amplification in breast invasive carcinomas, liver hepatocellular carcinoma, cholangiocarcinoma, uterine carcinosarcoma, thymoma, lung squamous cell carcinoma, lung adenocarcinoma, esophageal adenocarcinoma, and pancreatic adenocarcinoma, suggesting that PARP1 amplification could be an event of early stage in these tumors such as lung squamous cell carcinoma during cancer evolution (20), while endometrial carcinoma, skin cutaneous melanoma, stomach adenocarcinoma, adrenocortical carcinoma, bladder urothelial carcinoma, colorectal carcinoma, sarcoma, diffuse large-B cell lymphoma, and prostate adenocarcinoma may be easier to detect PARP1 mutations or deep deletions, which is somewhat supported by a prior study reported that multiple mutations typically predated amplification using molecular time analysis (20). Most advanced-stage tumors, including skin cancer, non-melanoma, endometrial cancer, small cell lung cancer, small bowel cancer, gastrointestinal neuroendocrine tumors, and bladder cancer, unlike early-stage tumors, are more prone to arising PARP1 mutations, and a study indicated that PARP1 point mutations were closely linked to de novo resistance to PARP1 inhibitors (8), which are usually regarded as synthetic lethal gents in homologous recombination (HR)-deficient tumors, and recent studies have also found that they can prolong progression-free survival in ovarian cancer patients with wild-type BRCA relapse (22, 23) or advanced ovarian cancer patients regardless of BRCA mutation (24), extending their application scope. Therefore, the selection of a suitable subsequent treatment strategy for patients with PARP1 alterations, and their prognostic value, remains unclear. We found the co-occurrence of genetic mutations in tumors with PARP1 alterations to be involved with some genes such as KMT2D/2C as driver genes, PIK3CA, ARID1A, H3F3A, HIST3H3, RYR2, and thus, PARP1 alteration should promote tumor mutability. Therefore, we analyzed the relationship between PARP1 alterations and TMB levels. Surprisingly, the MSK-IMPACT clinical sequencing cohort from the TCGA database (25) showed that TMB levels in the PARP1 altered group were significantly higher than those in the unaltered group, and similar results were verified in two ICI-treated cohorts (26, 27), which indicated that PARP1 alteration could predict ICI treatment effectiveness across tumor types because TMB level is commonly regarded as a notable biomarker associated with the treatment effect in many ICI-treated tumors (27).

We further evaluated the correlation between PARP1 alterations and the clinical prognosis. Patients with PARP1 alterations had markedly longer OS than those with PARP1 unaltered in the whole group and early-stage tumors, and a similar result was found in the ICI-treated cohort. Subgroup analysis revealed that the multiple PARP1 altered group had the best OS compared to those with a single PARP1 altered group. Notably, there was no association between PARP1 alterations and OS/PFS shown in microsatellite-stable (MSS) solid tumors, whereas compared to the PARP1 unaltered group, PARP1 alterations were somewhat linked to better ICIs treatment effect (42.9% vs. 27.7%). In early-stage tumors, the PARP1 altered group was closely associated with MSI. The evidence mentioned above suggests that PARP1 alterations might harm MMR and cause tumor phenotype to mutate, or even collaborate with ICI treatment; however, these findings remain to be validated in future studies.

To uncover the potential mechanism of PARP1 alterations that are linked to better ICI treatment effect, we analyzed the relationship between PARP1 alterations and immune infiltrates, including CD8+ T cells, CD4+ T cells, B cells, neutrophils, macrophages, and dendritic cells in all types of tumors from TCGA data by Timer2.0 (15, 28, 29). Impressively, PARP1 mutation was markedly associated with higher levels of immune infiltrates, such as CD8+ T cells, in several tumor types, suggesting that tumors with PARP1 mutations could be classified as active immune subtypes. Meanwhile, the PARP1 alteration group had significantly higher expression of immune checkpoint genes, including LAG3, CTLA4, PDCD1, and TIGHT. These data provide important evidence for PARP1 alterations as pan-cancer predictive biomarkers for ICI treatment. Current research found PARP1 inhibitors (PARPi) remodels tumor immune microenvironment, upregulates PDL1 and drives a systemic Th1-skewing immune response, which activates the priming of immunity and tumor-killing activity, in combination with ICIs for the renaissance of anti-tumor immunity (30), which is somewhat agreement with our results that some patients with PARP1 alterations have an immune-activated microenvironment by promoting tumor mutability, and could boost the sensitivity of patients to ICIs treatment, indicating PARPi and PARP1 alterations might possible represent the similar biological significance to some extent. In addition, although PARP1 alterations like mutations contribute to PARPi resistance (8), our results suggest these partial patients could benefit from ICIs immunotherapy, and another study reported that PARPi combining with chemotherapy, ICIs, alongside with targeted drug had the great advantage in overcoming PARPi resistance as well (31). Therefore, it is necessary to screen the patients with PARP1 alteration for selecting suitable therapeutic strategies.

Moreover, PARP1 alterations were positively associated with high transcription levels of PARP1, which might partially be due to increased PARP1 copy numbers, as evidenced by our results, which is in agreement with results of studies performed on breast cancer (32) and cervical cancer, respectively (33). However, the exact association between PARP1 alteration and its function remain largely unclear, our results showed that PAPR1 altered group, especially in the early-stage tumors, had a better prognosis and the opposite trend is present in advanced-stage tumors. Meanwhile, PAPR1 altered group in the early-stage tumors were predominance of its copy number amplification, and advanced-stage tumors seemed to be predominance of its mutations. Therefore, we speculated, based on those mentioned above, that PAPR1 copy number amplification and its mutations might lead to playing different role in the function of PARP1, thus caused the prognostic difference in early-stage and advanced-stage tumors. We will further explore and verify in future research.

A previous study found that overexpression of BRCA1-associated protein (BRAP) binding to breast cancer suppressor protein (BRCA1) was linked to worse prognosis and immune infiltration in several tumors, and used to as a potential molecular biomarker (34). Therefore, we speculated that DNA damage repair (DDR)-associated gene could have the important biological significance regarding clinical prognosis and treatment. Subsequently, we also analyzed the associations between PARP1 expression and clinical prognosis, immunotherapy signatures. We found that most tumors had high expression of PARP1 mRNA, compared to corresponding normal tissues, and there was a worse prognosis in patients with high expression for several tumors, suggesting that PAPR1 plays an oncogenic role to some extent in multiple tumors, which is consistent with the findings of previous studies on different tumors, such as colorectal cancer (35), gastric cancer (36) and sarcoma (37). Thus patients with high PARP1 expression are required to be monitored with closer follow-up protocols. Another possible reason for PARP1 overexpression is defective PARP1 cleavage, which leads to an imbalance of apoptosis induced by various chemotherapeutic drugs in tumor cells (38, 39), suggesting that these patients with high PARP1 expression might be responsible for chemoresistance. Therefore, whether immunotherapy is suitable for these patients remains to be investigated. We found that high PARP1 expression was significantly associated with six immune cells (B cells, CD4+ T cells, CD8+ T cells, macrophages, neutrophils, and dendritic cells) in most tumors, including COAD, HNSC, KIRC and LIHC etc., indicating that high PARP1 expression group had a greater number of immune cell infiltrations, which is agreement with high PARP1 expression being positively associated with higher immune score and stromal score in KIRC, COAD, UVM, KIRP and READ. In particular, CD8+T cell infiltration, which was also positively correlated with high PARP1 expression in BLCA, BRCA, KIRP, LGG, LIHC, PAAD, PCPG, PRAD, READ, TGCT, THYM, UCEC and UVM. Meanwhile, PARP1 expression was significantly positively correlated with the transcription levels of immune checkpoint genes, such as CD274, CTLA4, and PDCD1, in several tumors including PAAD, LIHC, KIRC, BLCA, and HNSC. These findings mentioned above are similar with a prior study showed that in some tumors, nuclear factor erythroid 2 like 2 (NFE2L2) that is upregulated by transcriptional activation of PARP1 was correlated with immune infiltration and also a potential prognostic biomarker (40, 41). Moreover, there were distinctly negative associations between PARP1 expression and HLA-I and II molecular levels in several tumors. Therefore, we speculated that immune escape may be involved in PARP1-mediated tumorigenesis; thus, immunotherapy such as anti-PDL1/PD1 and/or CTLA4 may be more susceptible in making them have benefits based on these results. Future studies on how to regulate HLA I and II molecular expression that is related to antigen presentation in several tumors with high PARP1 expression, need to be undertaken.

We further performed the analysis for the association of PARP1 expression and the number of immune neoantigens, and a significant positive association were showed between them in COAD, KIRC, LUAD, PAAD and THYM, simultaneously, there were also significant correlations between PARP1 expression and TMB in many tumors, some of which had positive associations including ACC, COAD, KICH, LGG, LUAD, READ, SKCM and STAD. In addition, high PARP1 expression was positively associated with microsatellite instability event in COAD, KIRP, BRCA, GBM, LUSC, LGG, READ, UCEC, SKCM and LUAD, indicating that these tumors with high PARP1 expression could have a better response rate to ICIs immunotherapy, in particular, for LUAD, COAD and READ, LGG with high PARP1 expression, which still needs to be verified in future studies.

In conclusion, the findings of the present study indicate that the pan-cancer analysis of PARP1 alterations and expression were correlated with clinical prognosis, and PARP1 alterations might act as biomarkers in the prediction of immunotherapy effects, and its expression levels seemed to be correlated with the status of immunotherapy-associated signatures, thus they may become promising biomarkers in the prediction of ICI response in several tumors. However, the limitation of this study lies in the need for further experimental verification.

Publicly available datasets were analyzed in this study. This data can be found here: All the data sets used in this study were publicly available at cBioPortal (https://www.cbioportal.org), TIMER2.0 (http://timer.comp-genomics.org/), and TCGA database (https://xenabrowser.net or https://portal.gdc.cancer.gov/).

XZ and JC designed this study. XZ and YW analyzed the data and wrote the manuscript. GA analyzed the data. JC revised the manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by the Medical Science Funding of Guangdong province (A2020139), the Youth Foundation of National Natural Science Foundation of China (81902420), and Guangdong Esophageal Cancer Institute Science and Technology Program Project (Q201903).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.