Secondary Cancer after Androgen Deprivation Therapy in Prostate Cancer: A Nationwide Study

Androgens play diverse roles in the body, encompassing the development of male secondary sex characteristics, erythropoiesis, body metabolism, bone density, and libido [1]. Normal androgen activity relies on androgens signaling through their receptor, the androgen receptor (AR), leading to the transcription of AR target genes.

Recent research has increasingly focused on androgens due to their involvement in various diseases, including metabolic syndrome, diabetes, cardiovascular disease, osteoporosis, dementia, and, notably, late onset hypogonadism in males [2]. This emphasis has also extended to oncological disorders, spurring a multitude of studies that investigate the role of AR expression in prostate cancer and other malignancies, such as breast cancer, lung cancer, and hepatobiliary cancers [3, 4]. Initial epidemiological observations revealed significant differences in cancer incidence and survival between sexes [5, 6], suggesting possible pathophysiological roles of androgens or AR expression, besides other sex hormones.

AR signaling is involved in various cancer types, similar to EGFR, HER2/Neu, AKT/mTOR/PI3K, and Wnt pathways [7, 8, 9, 10]. Traditionally, the central role of AR expression in the development and progression of prostate cancer is well-established [11]. Furthermore, the expression of AR target genes is highly relevant to androgen deprivation therapy (ADT) status [9]. Beyond prostate cancer, the involvement of AR in other malignancies has also been investigated, offering prospects for potential treatments involving antiandrogen therapies [12]. Recent studies also suggest that AR, as a prognostic marker in breast cancer survival, can affect estrogen receptor-dependent transcription [13, 14]. Consequently, strategies for treating breast cancer have shifted towards blocking AR signaling [15], with ongoing clinical trials, including phase 2 and phase 3 trials [4]. Investigations have hinted at a possible role of AR expression in cancer growth and drug resistance, not limited to prostate or breast cancer [3, 4]. However, few clinical studies have directly linked AR expression to cancer development, often due to challenges in accurately measuring AR expression and interpreting androgen activity based on serum testosterone levels.

The American Heart Association, in an observational study, linked ADT to an increased risk of cardiovascular events [16]. Therefore, researchers are now studying how ADT affects various diseases, with a particular focus on cardiovascular risks. Some new research suggests that ADT may harm the blood vessels by causing oxidative stress [17]. While ADT might impact cancer incidence by affecting cell growth and death through androgen signaling pathways, there isn't much ongoing research on this aspect.

Our study is based on the hypothesis that androgen signaling is associated with the development of secondary cancers, which holds promise for potential treatment using ADT in various malignancies. In this study, we investigated the pleiotropic role of androgen deprivation in different cancers, except for gynecological cancer.

Our findings indicate a positive association between ADT in the treatment of prostate cancer and the risk of developing brain and other CNS cancers, while revealing an inverse association with the risk of liver, colorectal, and lung cancers. To the best of our knowledge, this is the first clinical observation study suggesting that ADT could be a risk factor for various cancer types.

As the role of AR-driven transcriptional activities in prostate cancer has been well-established, recent studies have investigated its role in transcription and genomic stability across various cancers, with positive results [18]. However, the expression and activation of the AR in various cancers are not yet fully understood due to heterogeneous results attributable to clinical factors such as tumor type and stage [3, 19].

Recently, Hu et al [12] have described the role of AR in the development and outcome of various tumor types by pan-cancer analysis of the Cancer Genome Atlas (TCGA)-Clinical Data Resource. They reported that correlations of AR expression with incidence and survival varied across tumor types. AR expression was positively associated with stomach cancer and low-grade glioma. In contrast, it showed inverse associations with kidney cancer, acute myeloid leukemia, liver cancer, ovarian cancer, and melanoma [12]. Although this pan-cancer analysis was a precedent at the genomic level, its data have limitations for clinical applications due to inconsistent expression profiles between mRNA and protein levels. Moreover, the TCGA data lacked lung cancer data. Such data have been upgraded since 2014. Furthermore, interpretation of the TCGA clinical data is onerous due to challenges associated with analyzing and managing the complexity of huge data sets. Although we could not assess the expression at the genomic level, our findings represent the first real clinical data that could provide useful insight for future preclinical and clinical investigations regarding the role of AR in cancer therapy using ADT.

Ongoing clinical trials that are exploring AR-focused therapies for various cancers including hepatocellular cancer, bladder cancer, pancreatic cancer, AR-positive salivary cancer, renal cell cancer, and mantle cell lymphoma could provide further insight into the AR’s involvement in malignancies [3, 4]. Such clinical trials have been the basis of the hypothesis that AR expression could be associated with cancer incidence and aggravation not only in prostate cancer, but also in other cancer types. While our study couldn't directly establish this association, it did reveal intriguing patterns, including a potential inverse link with stomach cancer. AR expression has previously been associated with the incidence of liver cancer including hepatocellular carcinoma [6, 20], with several studies pointing to associations with worse clinical stages and poor outcomes [6, 21, 22]. Moreover, many studies have shown a positive association between AR signaling and increased risk of hepatitis B and C, known risk factors for hepatocellular cancer [23]. AR expression has been confirmed in both normal human lung cells and lung cancer cells [24, 25]. AR stimulation by androgens can stimulate the growth of small cell and non-small cell lung cancer lines in vitro [26]. Recently, Grant et al [27] have reported that androgen pathway blockade can significantly improve the survival of lung cancer patients.

Inconsistent findings between previous studies, clinical trials, and our observations suggest that there should be focus on the dual or complex action of AR depending on clinical cancer stage. These discrepancies could be due to small sampling sizes and different analysis protocols employed in previous studies. Most studies, including our current study, could not show clinical staging, tumor size, tumor complexity, or intervention protocols. Thus, such inconsistent findings draw the focus towards a possible dual function of AR according to clinical stage. However, our study could show cumulative risk of AR expression by ADT in secondary cancer risk.

The role of AR is different between early stages of cancer development and advanced progression stages in non-prostate cancers, especially in hepatocellular carcinoma and lung cancer, as implied by earlier evidence on this topic. Currently, there are no definitive results pointing to a role of AR as a cancer promoter, protective effector, or suppressor of cancer progression across cancer types and stages [3, 12]. Recently, Ma et al [21] have reported that AR expression is upregulated only in tumors smaller than 3 cm and that AR expression is absent in advanced and severe liver cancer tissues. They also demonstrated the importance of AR using an in vivo model of L-ARKO mice lacking hepatic AR treated with a carcinogen. The absence of hepatic AR resulted in poor survival, increased aggravation, and metastasis. It suggests that the key mechanism involves suppression of NF-κB and p38 phosphorylation. Regarding lung cancer, Nishio et al [28] have reported that patients with advanced or recurrent non-small cell lung cancer treated with an epidermal growth factor receptor tyrosine kinase inhibitor show lower androgen levels after treatment. This implies that a low androgen level could be a negative prognostic marker after treatment [28]. Grant et al [27] have also reported that high Ki67 and low AR expression are related to a higher risk of death and recurrence of non-small cell lung cancer.

Although our study showed an association between AR and secondary cancer development using ADT in a large cohort, several limitations should be taken into account for a better understanding of our results. First, our study did not directly measure serum testosterone level or AR expression. However, direct measurement of serum testosterone has been shown to be less reliable due to its wide range normal value and its dependence on the biological cycle [29]. Moreover, direct measurement of AR expression is expensive and complex. It is only possible in an experimental setting. However, our previous studies [30, 31] have shown that PSA with normal range could be a surrogate marker for serum testosterone using meta-analysis. It also could be used a predictive marker for subclinical and clinical cardiovascular diseases, indicative of androgen activity. Furthermore, our data could not present clinical staging, complicating objective interpretation, or relevance of clinical results. However, the clinical stage of prostate cancer itself could net affect the risk of secondary cancer development. This study population consisted of Korean men. It is well known that androgen activity is affected by ethnicity. Therefore, our findings might not be generalizable to other age groups, populations with a higher prevalence of comorbidities, or other race/ethnic groups. In addition, we could not explain the higher overall incidence of secondary cancer in the ADT group than in the non-ADT group. Lastly, despite adjustments for various medical service utilizations, this study, which used health insurance claims data, could not fully account for differences in follow-up intensity based on ADT status.

It is observed that ADT may be related to an increased or decreased risk of specific malignancies. The variations in risk suggest that the influence of ADT on different types of cancers is complex and multifaceted. While we must interpret these results cautiously, our study suggests that AR expression may play a role in cancer development and progression, with the impact varying based on the specific cancer type. In the future, rigorous randomized controlled trials (RCTs) and studies on mRNA and AR protein levels are crucial to understand the relationship between various ADT types and secondary cancer. Furthermore, the need for extended real-world evidence assessing the effects of ADT types on secondary cancers is paramount.

Our study demonstrated a positive association between the use of ADT in the treatment of prostate cancer and the risk of developing brain and other CNS cancers. Simultaneously, it revealed an inverse association between ADT and the risk of liver, colorectal, and lung cancers. This study supports the dual role of AR in various cancer types. Further studies including RCTs and genomic analysis are needed to elucidate the relationship between ADT and pan-cancer clinical features.

Supplementary materials can be found via https://doi.org/10.5534/wjmh.230237

Supplement Table 1

Cancer codes for analysis

Supplement Table 2

Pharmaceutical codes of study drugs

Supplement Table 3

Overall all cause of secondary cancer incidence and individual overall cancer-specific incidence before propensity score matching

Supplement Table 4

Individual cancer-specific incidence according to duration of androgen deprivation exposure before propensity score matching