Recurrent Profile of PC in Maranhão, Brazil
PC is a neglected and aggressive disease with a sizable impact on quality of life . Its incidence is high in developing countries such as Brazil, particularly in the North and Northeast regions, representing a serious local public health problem [1, 8]. In 2018, our research group found that the state of Maranhão in Northeast Brazil has the highest incidence of PC ever recorded in the world . In 2020, we characterized the sociodemographic features of patients with PC in the region .
In this study, we evaluated 200 PC cases diagnosed in Maranhão, integrating clinical-histopathological, immunohistochemical, and HPV infection for a better understanding of disease etiology. The cases were mostly composed of men in their 60 s, the age at which the risk of developing PC is higher; however, it still frequently occurs in younger men [1, 9, 21, 22].
Men under 60 years old accounted for 42.3% of all cases evaluated, with the youngest being only 23 years old. In two studies previously published by our group, 19.7% and 22% of the men were less than 40 years old, respectively [7, 8]. In the present study, we observed a decrease in the survival time of younger individuals (18–40 years old) in relation to those over 40 years old. According to the literature, younger men seem to develop mostly HPV-associated PC, being more prone to relapse, infiltrative patterns, and perineural invasion [23, 24].
We note the prevalence of individuals from the North of Maranhão, where the reference centers are located (in the capital, São Luís). The southern region seems to account for the lowest percentage of cases and this may be associated with the evasion of these patients to cities in neighboring states, such as Piauí (closer than to the capital), or to the southern region of Brazil (looking for better therapeutics approaches). The large migration of PC patients from poor regions to underdeveloped areas is commonly reported in Brazil , suggesting that the incidence of PC in Maranhão is possibly even higher than that described by Coelho et al. (2018) .
Socioeconomic vulnerability seems to be one of the main risk factors for PC . In the present study, the high prevalence of the disease in men with low education, those with a family income less than a Brazilian minimum monthly wage, and residents/workers from rural areas, reinforce this profile. However, it is important to remember that knowledge about PC and its risk factors seems to be scarce, even among highly educated men .
Exploring the profile of these patients, we observed that most of them reported poor genital hygiene habits, an important and known risk factor for PC . Proper penile hygiene has been one of the main prophylactic measures for PC, and when not performed, it favors the emergence of a favorable microenvironment for the proliferation of uropathogenic microorganisms, such as HPV . In addition, it increases the chances of the onset of episodes of chronic inflammation (balanitis and lichen sclerosus), which may also contribute to PC development [29,30,31].
In addition to the level of education, phimosis also seems to contribute to inadequate cleaning of the penis due to the difficulty or inability in exposing the glans. This condition was described in 70.5% of the cases evaluated in our study and is frequently reported in PC patients. In general, the prevalence of phimosis in PC is approximately 25% to 75%,  while it occurs only in 3.4% of healthy adult men . Some studies suggest that phimosis increases the risk of developing PC by 3.5 times, with no protective effect if dealt with after the neonatal stage [3, 34, 35].
Furthermore, we observed a high frequency of risky sexual behavior (54.1%), which can significantly contribute to the acquisition of sexually transmitted infections, such as HPV [36, 37]. A high prevalence (58.7%) of patients who had sex with animals (zoophilia) was shocking data but had been reported before. Vieira et al. (2020) described the practice of zoophilia in 60% of the cases in Maranhão . Both our data and those published by Vieira et al. (2020) corroborate the findings of Zequi et al. (2012), who demonstrated via univariate and multivariate analyses that zoophilia is an important risk factor for PC, but does not seem to impact pathological and clinical prognoses .
Upon evaluating the primary symptoms of these patients, we found that the medical records mainly described the appearance of a nodule/tumor in the penis (46.5%) or a foul-smelling ulcerative wound (21.0%). These symptoms are consistent with the main manifestations of the disease, which include changes in skin thickness or color, a wound not having healed in four weeks, rash, bleeding from the penis or under the foreskin, and a smelly odor . Vieira et al. (2020) described the presence of pruritus as the primary initial symptom in patients with PC in Maranhão (26.6%) and discussed its association with lichen sclerosus, despite only 6.0% of cases having this condition in their study . Interestingly, in our study, 27.3% of the cases had associated lichen sclerosus, and the presence of pruritus was observed in 20.2% of the cases, corroborating the findings of Nasca et al. (1999)  and Vieira et al. (2020) .
As the penis is an external organ, the PC symptoms are classic and easily recognizable. However, we noted an average delay of 17 months between the onset of symptoms and seeking medical care, similar to what was previously described in Maranhão by Vieira et al. (2020), who described a delay of 18.9 months . Late diagnosis is generally attributed to a lack of knowledge about the symptoms of the disease, associated stigmas, and fear of seeking medical help . These factors are intensified in Maranhão, where a large part of the population is socioeconomically vulnerable . The region has the second lowest Human Development Index (HDI) in Brazil (0.639), has rural socioeconomic characteristics, and has historically been marked by significant social inequality and extreme poverty .
In this context, access to basic health, especially in rural areas, is scarce, making it difficult to carry out follow-up and early diagnosis of several diseases, including PC. All these factors contribute to an advanced stage at diagnosis, a high rate of metastasis, limitations in the therapeutic approach, and a low overall survival rate. Population-based analyses from Europe and the United States have shown that, unlike other types of tumors, tumor-specific survival rates for PC have not shown any improvement since 1990 [43, 44].
Our findings reinforce a recurrent social-economic profile of patients with PC [7, 8, 45, 46] and call attention to an alarming situation of this disease in poor regions, especially in Maranhão [7, 8, 47]. We suggest that the sum of factors that simultaneously favor disease onsets, such as late diagnosis and advanced stages, may contribute significantly to the high incidence of PC in Maranhão, Brazil, and public policies aimed at remedying these factors in the region are urgent.
High frequency of HPV-associated PC in Maranhão: differences in disease behavior
The high prevalence of HPV-positive tumors reported in the last five years has been one of the most notable findings in Maranhão. The first study on the detection and genotyping of HPV in local cases was conducted by De Sousa et al. (2015)  and Martins et al. (2018) , where HPV was reported in 63.15% and 89.1% of tumors analyzed, respectively. Recently, another study described the presence of HPV in 96.4% of cases, the highest frequency ever recorded for PC .
All these studies carried out in the region showed a frequency of HPV-positive tumors above the global average of approximately 46.9—50.8% [49,50,51], and describe HR-HPV 16 as the most common viral type associated with PC. To better understand this situation, in the present study, we used morphological, immunohistochemical, and molecular criteria to investigate the involvement of HPV in the etiology of PC. Based on this, when we performed the molecular detection of HPV, we found a global frequency of HPV infection in 80.5% of PCs, mainly by HR-HPV (84.5%), especially type 16 (75%). We found an association between global HPV positivity and tumors located in the glans, those with carcinoma in situ, and pT1. In fact, glans has been reported as a frequent site of penile HPV infection (35.8%) and is more likely to be infected with multiple HPV types than other sites .
In DFS analysis, we did not observe any difference in survival curves according to the HPV status. In contrast, in a cohort from Maranhão, Martins et al. (2018) found better DFS in HPV-positive PC patients than in HPV-negative PC patients ; this result has already been reported for PC and other HPV-associated tumors, such as the head and neck [53,54,55]. Glegoire et al. (1995) demonstrated an association between HPV positivity and worse prognostic factors, such as high staging and aggressive growth patterns ; Wiener et al. (1992) found that the overall survival and DFS did not differ between groups in Kaplan–Meier analysis .
Histopathological analysis revealed morphological alterations suggestive of HPV infection (koilocytosis) in 78.5% of the evaluated cases. This corroborates previous findings in Maranhão, where koilocytosis was observed in 74.5% of the PC analyzed . The presence of koilocytosis was associated with men reporting more than 10 lifetime sexual partners, tumors with HPV-associated histology, and warty subtype. These findings illuminate the pathophysiology of HPV-associated PC and that in men with risky sexual behavior .
Koifman et al. (2011) observed koilocytosis in 88.4% of PC cases evaluated, with high specificity (90%), but low sensitivity in detecting the virus . De-Paula et al. (2007) observed the presence of koilocytosis in 63.1% of the PC samples evaluated, being associated with tumors with low or moderate staging and better DFS . In Maranhão, Nascimento et al. (2020) described angiolymphatic invasion and absence of koilocytosis as predictive factors for lymph node metastasis , corroborating previous findings by Ornellas et al. (2008) . However, in the present study, we found no difference in DFS between the groups with and without koilocytosis, as well as no association between koilocytosis and lymph node metastasis.
The presence of koilocytosis is well accepted as pathognomonic evidence for possible the presence of HPV, and has been widely used in screening for cases of viral infection . However, koilocytes are not always present; therefore, it is necessary to use more sensitive and specific methods that also provide information about the viral type and oncogenic risk .
To better understand aspects related to HPV infection, we searched the literature for data on the global prevalence of the virus in the general population. In Brazil, the prevalence of HPV in individuals without clinical symptoms appears to vary from 30 to 75%, depending on the region studied . Interestingly, Maranhão represents one of the highest rates recorded in the country (59.1%), according to sampling carried out in the capital (São Luís) . It is believed that in the interior regions of the state, where most PC patients come from, these rates may be even higher, but there are no studies with a representative sample of these locations.
The high prevalence of healthy individuals infected with HPV in Maranhão has been attributed to factors such as early initiation of sexual activity (average age, 15.5) and the high number of sexual partners . In the present study, we did not assess the onset of sexual activity, but we observed a statistically significant association between risky sexual behavior and the presence of koilocytosis in PC patients, which corroborates the epidemiological HPV aspects of this disease.
Knowledge of the regional prevalence of HPV infection is important for understanding the local etiology of PC, as the incidence of HPV-associated tumors seems to be proportional to the prevalence of viral infection in healthy individuals . In our series, when we searched for tumors that were negative for koilocytosis, HPV infection, and p16INK4a overexpression, only one case was not positive for any of these three parameters. Interestingly, this was the first case report of pseudoangiosarcomatous squamous cell carcinoma on the penis, an uncommon and highly aggressive variant of squamous cell carcinoma .
In addition to koilocyte and HPV molecular detection, we performed p16INK4a protein expression analysis, which is widely used as an indirect marker of HR-HPV infection and has been associated with basaloid and high-grade tumors [9, 68, 69]. However, the absence of p16INK4a overexpression does not exclude infection by LR-HPV or even the absence of HPV infection [17, 70].
We found p16INK4a overexpression in 26.0% of PC cases. In the present study, all basaloid tumors showed p16INK4a overexpression, reinforcing its secondary use as a tool in the differential diagnosis of this histological subtype [9, 17, 70]. Furthermore, the lack of p16INK4a overexpression was associated with the absence of angiolymphatic invasion.
According to Cubilla et al. (2011), p16INK4a overexpression is closely correlated with the presence of the HR-HPV genotype in PC, as observed in cervical carcinoma and other HPV-related tumors [17, 70]. However, in the present study, no association between p16INK4a overexpression and HR-HPV infection was observed. We suggest that limitations in HPV detection methods may explain these findings, such as: (1) we used DNA samples obtained from FFPE tissue, unlike Martins et al. (2018), who used fresh tumor samples; and (2) viral genotyping was not done for all positive cases.
Despite the high frequency of cases with koilocytes and HPV-positivity, the frequency of p16INK4a overexpression was low compared to that described by Martins et al. (2018) in Maranhão, where 40% of the cases had overexpression of p16INK4a . We speculate that the use of different antibody clones for IHC of p16INK4a might underlie this difference. Martins et al. (2018) described four basaloid tumors in a total of 55 cases, while here we had eight basaloid tumors in approximately 173 cases.
We also did not observe a statistically significant difference in DFS with the level of p16INK4a expression. However, patients with p16INK4aoverexpression lived on average less (30.1 months) when compared to patients without p16INK4a overexpression (46.8 months). These findings do not corroborate with previous studies where PC men with HPV or p16INK4a overexpression generally have a significantly more favorable DFS compared to those without HPV or p16INK4a overexpression . Furthermore, a recent meta-analysis showed that p16INK4a overexpression was significantly associated with and was an independent factor for better DFS in PC patients (HR = 0.45, 95% CI: 0.30–0.67, P < 0.001) .
Given these controversial data, we performed DFS analysis for p16INK4a overexpression according to the following criteria: (1) basaloid tumors were excluded; and only (2) HR-HPV positive cases; (3) usual subtype tumors; (4) warty subtype tumors; and (5) stage I-II and III-IV tumors. However, in all analyses, there was no significant difference in DFS, and the mean survival time of the p16INK4a overexpressing group remained lower than that of no overexpressed p16INK4a.
Unfortunately, Martins et al. (2018) did not describe survival data based on p16INK4a expression, making it difficult to compare their findings in Maranhão with those observed in the present study . However, similar to what was observed in our study, in a study conducted in Northern Brazil (state of Amazonas), no differences were observed between the survival curves of PC patients with p16INK4a overexpression (log-rank p = 0.753) or molecular HPV positivity (log-rank p = 0.979) . According to Ornellas & Ornellas (2018), it is uncertain whether PCs involving HPV infection have better survival profiles than those without HPV infection . Our data and those published in Amazonas suggest a different impact of p16INK4a overexpression in PC patients from these Brazilian regions.
For histopathological classification, all the evaluated tumors were classified as squamous cell carcinoma (SCC), the most common type of PC . According to the WHO , these tumors are grouped into HPV-associated subtypes (basaloid, papillary-basaloid, warty, warty-basaloid, and clear cell carcinoma) and non-associated HPV (usual, pseudohyperplastic, pseudoglandular, verrucous pure or cuniculatum, papillary, adenosquamous, and sarcomatoid).
In the present study, most cases were the usual subtype (39.5%), followed by the warty (29.8%). These frequencies differ from other studies in the area, such as that carried out by Cubilla et al. (2001), where out of 61 tumors evaluated, 59% corresponded to the usual subtype and only 10% corresponded to warty tumors . In 2010, Chaux et al. compared the distribution of different histological subtypes of PC from regions of high (Paraguay, 144 cases) and low (USA, 157 cases) incidence and observed that the prevalence of histological subtypes in the Paraguayan and American series was as follow: usual, 49.3% and 46.5%; verrucous, 8.3% and 7.6%; papillary NOS, 7.6% and 5.7%; warty, 6.9% and 8.3%; basaloid, 4.2% and 7.0%; sarcomatoid, 0.7% and 0.6%; adenosquamous, 3.5% and 0.6%; and mixed, 19.4% and 23.6%, respectively .
In 2018, Martins et al. observed that 47.3% of the PC cases evaluated in Maranhão were usual and 29.1% were warty subtypes . Subsequently, Vieira et al. observed 40.0% of usual tumors and 33.0% of warty tumors . Another important aspect is the high frequency of mixed tumors (with characteristics of more than one histological subtype), which in the present study was approximately 32.0%, higher than that described by Martins et al. 2018 (16.3%) and Vieira et al. 2020 (25.0%). Considering mixed tumors according to the presence or absence of an HPV-associated component, similar to the classification described by Canete-Portillo et al. (2020) , all mixed tumors in the present study had a basaloid, warty-basaloid, or warty component associated with another component (usual). Thus, the overall prevalence of HPV-associated histology in the cases evaluated in our study go up to 77.5%.
The high frequency of warty tumors in the present study and in others carried out in Maranhão, as well as the prevalence of mixed tumors with HPV subtypes associated, differs drastically from that observed in other studies, and supports the high prevalence of HPV-associated PC in the region [76, 77].
Expression of p53 is Related to Worse Prognosis and Lymph node metastasis
When we evaluated the p53 expression profile, negative cases were more prevalent (55.0%) than positive cases (45.0%). The statistical analysis revealed that p53 positivity was associated with worse prognostic factors such as high-grade tumors (G3), pT3-pT4, those in stage II, and those with lymph node metastasis, corroborating previous studies with PC and other tumor types [79,80,81].
Previously, all poorly differentiated PCs analyzed by Marinescu et al. (2016) were p53 positive , and Lopes et al. (2002) found that p53 expression was an independent factor for lymph node metastasis (relative risk 4.8, 95% CI 1.6 to 14.9); the 5 and 10-year overall survival rates were 65.2% and 54.6% in PCs with negative p53, and 30.2% and 26.4%, respectively, in those with positive p53 . In this study, p53 expression was associated with lower DFS when compared to negative cases.
According to the DGIdb database, the TP53 gene is the target of 193 therapeutic drugs , many of which are already used for head and neck tumors, which appear to have a mutational signature very similar to PC . Overexpression of p53 is frequently associated with disruptive mutations and oncogenic activation in HPV-negative PC, possibly in cooperation with other mutations in genes of the HER-PTEN-AKT pathway .
In a study conducted by Jacob et al. (2019), most patients with metastatic PC had mutations in TP53 and were HPV-negative  This suggests that TP53 mutations appear to be the main factor associated with metastasis in patients with advanced PC, whereas HPV-related oncogenic mechanisms appear to be more effective in the early stages of the disease and are gradually lost [85, 86]. These findings may explain the pathophysiological differences observed in this study and in other studies carried out in Maranhão, where most cases are diagnosed in the advanced stages of the disease.
In contrast to what we observed for p53 positivity, the absence of p53 expression was associated with warty tumors, presence of carcinoma in situ, tumors with HPV-associated histology, absence of perineural invasion, koilocytosis, and expansive growth pattern. These findings corroborate the main pathway for HPV action, where the viral oncoprotein E6 binds to the p53 protein and inhibits its tumor suppressor function . In addition, a previous study in Maranhão found that 81.0% of PC cases were negative for p53 expression by IHC and 83.3% presented TP53 gene downregulation .
Ki-67 Expression as a Marker of Tumor Aggressiveness in PC, Including Perineural and Angiolymphatic Invasion
Most tumors analyzed in the present study had a low global expression of ki-67 (73.4%), but high expression in hotspot zones (57.8%). Here, we observed an association between high ki-67 expression and basaloid and high-grade tumors, presence of angiolymphatic and perineural invasion, pT3-pT4 tumors, stage III-IV, and lymph node metastasis. The association between high expression of ki-67 and worse prognostic factors has been previously described in PC [19, 88, 89].
Ki-67 analysis has been used as a good parameter to assess the proliferative index and aggressiveness of numerous tumors [90, 91]. In PC, Stankiewicz et al. (2012) found an association between high expression of ki-67, basaloid tumors, and HPV infection, corroborating the results of the present study, in which all basaloid tumors had high ki-67 expression . In contrast, we observed that warty tumors were associated with low ki-67 expression (global and hotspot). This difference corroborates the clinical behavior of each histological subtype, where basaloid tumors generally have fast growth and generally poor prognosis, while warty tumors have slow growth and are more often a good prognosis .
A strong association between high ki-67 expression and poorly differentiated tumors has been described previously in PC . Berdjis et al. (2005) observed an inverse relationship between ki-67 expression and tumor differentiation (p < 0.0005), but they did not observe a relationship with advanced staging or lymph node metastasis, unlike our findings .
The association between high ki-67 expression and angiolymphatic and perineural invasion is another important finding of our study and was previously described for PC. Angiolymphatic and perineural invasions are known to be associated with a worse prognosis, and its detection is sometimes influenced by the observer (pathologist) bias, suggesting that IHC for ki-67 can refine the diagnosis and prompt one to search for neural or vascular invasion in the histopathological analysis [92, 93].
Protzel et al. (2007) and Zhu et al. (2007) described a strong association between high ki-67 expression and lymph node metastasis in PC [79, 88], and Warli et al. (2020) argued that this association appears to be independent of grade and stage . Our findings also demonstrated an association between high ki-67 expression and the occurrence of lymph node metastasis in PC patients, indicating the role of ki-67 as a predictive factor for metastasis in these patients.
Despite the association with worse prognostic factors and lymph node metastasis, we did not observe a statistically significant difference between the DFS rate according to the ki-67 expression profile, similar to the study published by Stankiewicz et al. (2012) . We hypothesized that the cumulative effect of other factors such as staging, angiolymphatic invasion, and perineural invasion in our study may have influenced DFS patients more than ki-67 high expression. For our studies, future multivariate analyses are required to elucidate these aspects. However, high ki-67 expression has been reported to be associated with worse DFS in PC (log-rank p = 0.0098) .
Considerations and perspectives
PC is a neglected, heterogeneous, aggressive, and deadly disease that is typically not diagnosed early. There is a high incidence of PC in the Maranhão state, and little is known about the local etiological profile of these tumors. Most patients are socioeconomically vulnerable and come from the state's countryside. There are few centers for treating the disease in the region, most of which are located in the capital, São Luís, making early diagnosis and monitoring of these patients challenging.
Here, we discussed the impact of the geographic isolation of the state capital on the displacement of these patients, suggesting that the actual incidence of the disease is more significant than that described previously . Our findings reinforce a profile of the disease in the state consistent with other studies and bring new insights into possible factors that, together, contribute to Maranhão having the highest incidence of PC in the world, such as the long delay in diagnosis and the high prevalence of HPV infection.
According to Ornellas & Ornellas (2018), when PC is considered globally, only a tiny proportion is described as HPV-associated. Therefore, increased men's education and prevention strategies such as condom use, hygienic measures, smoking cessation, and avoidance of chronic inflammatory states can considerably impact the pathogenesis of precancerous lesions of the penis . In Maranhão, where a significant number of PCs appear to be HPV-associated and most patients are socioeconomically vulnerable, both HPV immunization and patient education could impact disease rates.
Our data discuss HPV infection in a multi-methodological way, integrating histological, molecular, and immunohistochemical aspects and bringing new insights into the role of HPV on histological presentation and clinical behavior of these tumors. An interesting question involves the histological and prognostic variation observed in HPV-associated tumors. For example, why do warty and basaloid tumors, both HPV-associated, have widely different prognoses? Why does p16INK4a overexpression appear to be related to decreased disease-free survival and what is its impact on overall and cancer-specific survival in PC patients? To answer these questions we need to explore the correlation between CDKN2a (p16INK4a) copy number, mRNA expression, methylation profile, and protein expression (IHC).
Another important approach that we are currently investigating is the biological profile of viral oncoproteins (E6 and E7), both by gene (mRNA) and protein expression. These data linked to the characterization of p16INK4a can reveal the real status of HPV infection in these patients, whether it is active and impacting tumor progression or part of the HPV-positive cases are just a background due to the high rate of HPV infection in the general population, especially in Maranhão (as discussed).
The joint analysis of this information can help to better characterize each of the cited markers and their real role in the patient's prognosis. The perspective of our group has been, given the vast majority of HPV-positive cases, to investigate the differences within this group. We are currently performing all these analyzes on a prospective cohort of 60 cases from Maranhão, involving histological (H&E, IHC), genomic (exome, RNA-seq, miRNA-seq), and metagenomic tools to unravel these questions. All these approaches ongoing rely on the expertise of geneticists, physicians, pathologists, and oncologists with experience in PC to perform an integrative analysis between genomic (multi-omics) and the clinical aspects of PC. We hope with this approach to also elucidate the role of changes in p53 in PC carcinogenesis, evaluating somatic mutations and gene expression patterns.
At the time, the associations found between p16INK4a, p53, and ki-67 expression have clinical relevance and seem to impact the DFS of these patients. The test used to assess these protein expression profiles (IHC) is offered free of charge by the Brazilian Unified Health System (SUS) and can be used in histopathological routines to help diagnostic and prognostic classification of these patients.
Despite the difficulties inherent in the PC study, the present study strongly points to a clinical-histopathological and etiological profile different from that observed in other studies in Brazil and worldwide. It highlights the importance of local and collaborative research aimed at understanding these tumors' etiological aspects, behavior, and progression, especially in places with high incidences, such as Brazil. We emphasize that, due to limitations in current protocols for the management of PC patients, early diagnosis becomes even more decisive for better therapeutic conduct. This reality is expected to change since recently, the Brazilian consensus statement for managing PC patients in low- and middle-income countries has been proposed .
In the present study, we presented important clinical-histopathological and pathophysiological features of PC. However, some limitations may reduce the clinical impact of the data obtained and must be considered. First, the retrospective aspect of the study makes it difficult to have reliable control over the variables evaluated, creating gaps in the data. Second, the use of DNA from FFPE tumor samples may increase HPV-PCR false negatives and the absence of genotyping data for all HPV-positive tumors limits the stratification of cases according to the HPV genotype. Third, this study had many censored patients in DFS analysis due to loss of follow-up and there is no data regarding cancer-specific and/or overall survival. Despite this, the data obtained from a large Brazilian cohort of PC present a wealth of information that significantly contributes to a better understanding of this neglected disease. All limitations may be further explored in a future prospective study, as discussed.