The prognostic value of lymph node ratio in comparison to positive lymph node count in penile squamous cell carcinoma

Purpose Penile cancer is a rare male neoplasm with a wide variation in its global incidence. In this study, the prognostic value of lymph node ratio (LNR) was compared to that of positive lymph node count (PLNC) in penile squamous cell carcinoma. Methods A total of 249 patients with penile squamous cell carcinoma were enrolled from The Surveillance, Epidemiology, and End Results (SEER) database between 2010 and 2015. The X-tile program was used to calculate the optimal cut-off values of LNR and PLNC that discriminate survival. We used the χ2 or the Fisher exact probability test to assess the association between clinical-pathological characteristics and LNR or PLNC. Univariate and multivariate Cox regression analyses were performed to identify independent prognostic factors for survival. Spearman correlation analysis was used to determine the correlation between LNR and PLNC. Results We found that patients with high LNR tended to have advanced N stage, the 7th AJCC stage, and higher pathological grade, while patients with high PLNC had advanced N stage and the 7th AJCC stage. Univariate Cox regression analysis revealed that the N stage, M stage, the 7th AJCC stage, lymph-vascular invasion, LNR, and PLNC were significantly associated with prognosis. Multivariate Cox regression analysis demonstrated that LNR rather than PLNC was an independent prognostic factor for cancer-specific survival. Subgroup analysis of node-positive patients showed that LNR was associated with CSS, while PLNC was not. Conclusion LNR was a better predictor for long-term prognosis than PLNC in patients with penile squamous cell carcinoma.

Therefore, this study aimed to compare the prognostic impact of PLNC versus LNR in PSCC patients.

Patients and variables
A total of 249 men diagnosed with PSCC between 2010 and 2015 were retrospectively identified using the SEER*Stat software program. The inclusion criteria was as follows: (1) AJCC stage: Unknown; (2) SEER cause-specific death classification: NA/Unknown; (3) Regional nodes examined: 0-1 OR Unknown; (4) Regional nodes positive: Unknown; (5) Grade: Unknown. The screening process is as presented in Fig. 1.
The following variables were assessed: age, marital status, the 7th AJCC/TNM stages, histology, grade, primary site, lymph-vascular invasion, PLNC, and LNR. The age was grouped by patients' median age at diagnosis. Detailed information is as shown in Table 1. The endpoints of this study were overall survival (OS) and cancer-specific

Statistical analysis
The lymph node ratio was calculated as the ratio of the number of positive lymph nodes to the total number of lymph nodes examined. The optimal cut-off value was determined using the X-tile program. For OS, the optimal cut-off value of LNR was 0.23, with values ≤ 0.23 considered low and values > 0.23 considered high. The optimal cut-off value of PLNC was 3, with values ≤ 3 regarded as low while those > 3 were high. For CSS, the optimal cutoff value of LNR was identical to OS. The optimal cut-off value of PLNC was 1, with PLNC ≤ 1 considered low, and PLNC > 1 high. The χ2 test or the Fisher exact probability test was used to assess the association between the clinical-pathological characteristics and LNR or PLNC. Kaplan-Meier method was used to determine the survival analysis and the Log-rank test was used to examine the statistical differences between LNR or PLNC groups in terms of overall survival and cancer-specific survival. Spearman correlation analysis was used to determine the correlation between LNR and PLNC. Cox regression analysis was used to compute the hazard ratios (HRs) and 95% confidence intervals (95%CIs) for the identification of the prognostic factors in the survival of PSCC patients. All statistical analyses were performed using IBM SPSS Statistics 25. P values < 0.05 were considered statistically significant in the χ 2 test, the Fisher exact probability test, the Logrank test, and multivariate Cox regression analysis, while P values < 0.1 were considered significant in univariate Cox regression analysis.

The relationship between LNR/PLNC and clinical pathological characteristics in patients with PSCC
We used the χ 2 test or the Fisher exact probability test to compare the characteristics between the LNR/PLNC groups and clinical-pathological characteristics of PSCC patients. Our results revealed that high LNR patients tended to have advanced N stage (P < 0.001), the 7th AJCC stage (P < 0.001), and higher pathological grade (P = 0.048) while no significant association was found with other characteristics. PLNC was significantly associated with the N stage (P < 0.001) and the 7th AJCC stage (P < 0.001) based on the cut-off value of 1 and 3, while a significant association was observed between the M stage and PLNC grouping based on the cut-off value of 1 (P = 0.010) ( Table 2). These findings suggested that both high LNR and PLNC were associated with poor clinical-pathological characteristics in PSCC. Consequently, LNR and PLNC can serve as potential prognostic factors guiding clinical decisions.

The prognostic value of LNR and PLNC for survival in patients with PSCC
To further explore the role of LNR and PLNC in predicting the survival of PSCC patients, Kaplan-Meier analysis and the Log-rank test were used to estimate the overall survival and cancer-specific survival based on the LNR and PLNC status. There were significant differences in overall survival analysis between the LNR ( Fig. 2A) and PLNC groups (Fig. 2B). Patients with LNR ≤ 0.23 had a significantly higher 5-year overall survival rate than those with LNR > 0.23 (67.5 vs. 27.3%). The 5-year overall survival rate of patients with PLNC ≤ 3 and > 3 were 67.4 and 25.8%, respectively. A similar trend was observed in cancer-specific survival analysis ( Fig. 2C and Fig. 2D). The 5-year LNR lymph node ratio, PLNC positive lymph node count cancer-specific survival rate of patients with LNR ≤ 0.23 was higher than LNR > 0.23 (76.9 vs. 36.4%), whereas the 5-year cancer-specific survival rate of patients with PLNC ≤ 1 and > 1 were 81.7 and 51.1%, respectively. Even for the P values < 0.001, the χ 2 values of LNR were higher than PLNC, indicating that LNR may be a more promising prognostic factor for PSCC patients.
Considering that the lymph nodes harvested from lymphadenectomy comprise positive and negative lymph nodes, PLNC is theoretically correlated with LNR. Thus, we analyzed the correlation between LNR and PLNC by performing a Spearman correlation analysis. The results (rs = 0.926, P < 0.001) suggested that LNR and PLNC were significantly correlated (Fig. 3).
Next, univariate and multivariate Cox regression analyses were performed to investigate the independent prognostic factors influencing overall survival and cancer-specific survival. N stage, M stage, the 7th AJCC stage, lymph-vascular invasion, LNR, and PLNC (all P < 0.1) were found to have a significant impact on both overall survival and cancer-specific survival while age and T stage (both P < 0.1) only influenced overall survival (Table 3). Interestingly, no prognostic significance of pathological grade in PSCC was found, which was inconsistent with our general cognition of malignant tumors. A possible reason for this inconsistency is the limited sample size, hence the lack of adequate representation of the population.
Multivariate Cox regression models for survival were used to compare the effects of LNR and PLNC. Considering that the 7th AJCC stage contains information on the N stage and M stage, several multivariate Cox regression models incorporating lymph-vascular invasion, the 7th AJCC stage, and LNR/PLNC were constructed (  1.050, 1.224), P = 0.001) were found to be significantly associated with overall survival and cancer-specific survival (Table 4).

Subgroup and survival analysis in node-positive patients
Subgroup analysis was carried out to assess the association between clinical factors and survival in 132 patients with positive lymph nodes (Tables 5 and 6). Univariate analysis revealed that the N stage, M stage, the 7th AJCC stage, lymph-vascular invasion, LNR, and PLNC (all P < 0.1) were associated with both overall survival and cancer-specific survival whereas age and T stage (both P < 0.1) were only associated with overall survival. Multivariate analysis demonstrated that, LNR was an independent prognostic factor for both overall survival and cancer-specific survival (Model 1: HR = 2.612, 95% CI = (1.529, 4.461), P < 0.001; Model 2: HR = 2.994, 95% CI = (1.647, 5.440), P < 0.001) while PLNC (Model 4: HR = 1.447, 95% CI = (0.645, 3.248), P = 0.370) was not significantly associated with cancer-specific survival. These results suggested that LNR exhibited better prognostic value compared with PLNC in node-positive patients.
However, the absence of the 7th AJCC stage in all multivariate Cox proportional hazards regression models indicated that there were defects in our models. Large-scale analysis of complete and representative patients' information is needed for the calibration of the Cox regression model. In conclusion, LNR exhibited a better prognostic prediction than PLNC and could thus, serve as a promising prognostic factor in PSCC.

Discussion
Penile cancer is a relatively rare disease worldwide, it accounts for only 1% of all male malignancies, but causes considerable psychological and physiological trauma [1]. Squamous cell carcinoma is the most common histological type of penile cancer, with approximately 80% of cases localized in the glans penis and prepuce [13]. The diagnosis, treatment, and prognosis of PSCC are greatly correlated with lymph node status. The diagnosis of PSCC is mainly based on physical examination and regional lymph nodes evaluation [14]. Surgical resection with regional lymph node dissection remains the standard therapeutic modality for locally advanced cases [15] and an important prognostic values [16,17]. However, apart from metastasis, lymphadenopathy may also be caused by infection [18]. Thus, antibiotic treatment can prevent unnecessary lymph node biopsy. Besides, the lymphatic nodal metastasis status is the most significant prognostic factor in patients with penile squamous cell carcinoma [19]. Although men with less severe disease exhibit prolonged survival, the prognosis of advanced or metastatic PSCC remains poor, thus requiring a more robust prognostic index than the traditional AJCC TNM staging system. LNR and PLNC have exhibited prognostic value in a variety of tumors, including salivary gland cancer [20], prostate cancer [21], non-small cell lung cancer [22,23], breast cancer [24], and colon cancer [25]. To date, the association between LNR/PLNC and survival in patients with PSCC has not been well elucidated. Svatek et al. conducted a survey with 45 patients between 1979 and 2007 and reported that LNR (≤ 6.7 vs. > 6.7%) was significantly associated with CSS in patients with node-positive PSCC [26]. However, the small population and excessively long period limited the validity of the findings. Similarly, Lughezzani et al. observed significant differences in survival rates based on LNR [27]. The highlighted studies stratified survival outcomes using the median value, which is not very rigorous in determining the threshold. Besides, both studies did not compare the prognostic value of LNR and PLNC.
A recent study proposed that LNR was a better prognostic indicator compared to PLNC [28], however, the study population comprised only 28 penile cancer patients. Another study suggested that LNR, but not PLNC was a predictor of dismal survival outcomes in node-positive PSCC [29]. However, the study simultaneously added LNR and PLNC into the multivariate Cox regression model, which led to multicollinearity.
Compared with previous studies, the present study had a relatively large sample size, thus making it more applicable in clinical practice. Furthermore, there is still a considerable discrepancy in the threshold of LNR to discriminate between favorable and poor survival in prior studies, ranging between 0.067 and 0.33 [26][27][28][29][30]. This discrepancy may be explained by the use of varied surgical approaches, the extent of lymph node dissection, and the use of different statistical methods to calculate the optimal cut-off values. Unlike in previous studies where stratification of patients was done using the median, the optimal cut-off values for LNR/PLNC were determined using the X-tile program in this study. The optimal cut-off points were 0.23 for LNR (for both OS and CSS) and 3 (for OS)/1 (for CSS) for PLNC. When considered as categorical variables in both univariate and multivariate Cox regression analyses, LNR > 0.23 predicted worse survival outcomes compared with LNR ≤ 0.23 in patients with penile cancer. However, PLNC did not exhibit prognostic value in multivariate model predicting CSS. Thus, LNR was found to be a better prognostic factor for PSCC. Surprisingly, when converted to continuous variables, both LNR and PLNC were found to be independent prognostic factors of poor survival in penile cancer. The varying prognostic value exhibited by PLNC may be attributed to the loss of information contained in the raw data when converted to a categorical variable. However, a definite threshold is required by clinicians assessing prognosis in clinical practice. Therefore, investigating the prognostic value of LNR and PLNC as categorical variables makes it easier for clinical decision-making. Considering that heterogeneity may exist in patients with and without lymph node metastasis, subgroup analysis was performed in node-positive patients. LNR (P < 0.001) was found to be a better prognostic marker than PLNC (P = 0.370) for CSS. Accordingly, LNR was found to be a better predictor for survival than PLNC in patients with PSCC, and could also be used to distinguish between postoperative PSCC patients with poor prognosis requiring adjuvant therapy. The optimal cut-off value of 0.23 for LNR may not be directly used in clinical practice, and further consideration in combination with clinical information is needed.
The prognostic value of PLNC depends to a great extent on surgical and pathological procedures. In conditions of inadequate lymph node dissection, this can lead to the phenomenon of "stage migration" [31]. The superior prognostic value of LNR can be explained by the incorporation of disease burden and quality pathologic examination (PLNC), and the extent of lymphadenectomy (the number of examined nodes), which would reduce bias due to insufficient lymph node evaluation [32,33]. Thus, in the case of sufficient lymph node retrieval, the prognostic value of LNR may CI confidence interval, LNR lymph node ratio, PLNC positive lymph node count *Two-sided P value < 0.05 Despite the advantages of the present study, there were several potential limitations. First, this was a retrospective study based on a public database, and the study population was highly selected, which may result in selection bias. Second, the data obtained from the SEER database involved multiple centers. Thus, standardization of the surgical approach, especially the extent of lymphadenectomy, could not be implemented due to the multicentre nature of the study. Third, several important variables were not included in the SEER database, such as the size of lymph nodes, the region of lymph node metastasis (inguinal/pelvic lymph nodes), extranodal extension, tumor recurrence, and adjuvant therapy. These confounding factors influence survival but could not adjusted in our study. Finally, some common prognostic factors such as T stage and pathological grade were not associated with CSS in this study, possibly because the sample size of some stratified patients was too small.

Conclusion
In summary, we demonstrated that LNR is associated with the long-term survival of postoperative PSCC patients and is a better prognostic marker than PLNC. Besides, LNR can be used to stratify patients for adjuvant therapy in the case of inadequate lymph node dissection. Funding Not applicable.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
Code availability Not applicable.

Conflict of interest The authors have no conflicts of interest.
Ethical approval Due to the deidentified nature of data in SEER, the study was exempt from ethics committee approval. The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study conformed to the provisions of the Declaration of Helsinki (as revised in 2013).

Consent for publication Not applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.