Translate this page into:
Clinicodemographic and anatomical predictors of squamous cell carcinoma
*Corresponding author: Fatih İçbudak, Department of Plastic, Reconstructive and Aesthetic Surgery, Prof. Dr. Cemil Taşçıoğlu City Hospital, İstanbul, Turkey. icbudakfatih@gmail.com
-
Received: ,
Accepted: ,
How to cite this article: İçbudak F, Saridal MK. Clinicodemographic and anatomical predictors of squamous cell carcinoma. J Cutan Aesthet Surg. doi: 10.25259/JCAS_103_2026
Abstract
Objectives:
To evaluate the clinicodemographic characteristics and anatomical distribution of basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) and identify independent predictors of SCC.
Material and Methods:
A retrospective analysis of 374 patients (BCC = 209, SCC = 165) was performed. Variables included age, sex, and tumor localization. High-risk locations were the lip, scalp, and temporal regions. Univariate and multivariate logistic regression analyses were performed. Age was analyzed continuously and categorically (<60, 60–80, >80 years). Model performance was assessed using receiver operating characteristic analysis.
Results:
The mean age was 68.2 ± 12.8 years in the BCC group and 70.5 ± 12.7 years in the SCC group (p = 0.082). SCC was more frequent in males (72.1% vs. 56.0%, p = 0.003). In multivariate analysis, male sex (odds ratio [OR]: 1.87, p = 0.012) and high-risk localization (OR: 3.30, p < 0.001) were independent predictors of SCC. Age ≥80 years was also associated with SCC (OR: 2.54, p = 0.008), whereas age 60–80 years was not (p = 0.352). The optimal probability cutoff for SCC was 0.608 (sensitivity: 41.3%, specificity: 89.7%; area under the receiver operating characteristic curve [AUC] = 0.660).
Conclusion:
Male sex and high-risk anatomical localization are strong predictors of SCC. Age exhibits a nonlinear association, with increased SCC risk only in patients >80 years.
Keywords
Anatomical distribution
Basal cell carcinoma
Logistic regression
Non-melanoma skin cancer
Risk factors
Squamous cell carcinoma
INTRODUCTION
Nonmelanoma skin cancers (NMSCs), primarily basal cell carcinoma (BCC) and squamous cell carcinoma (SCC), are among the most frequently diagnosed malignancies worldwide, and their incidence continues to rise progressively.1,2 Although BCC is more common and generally indolent, SCC carries a higher potential for local invasion and metastasis, making its differentiation clinically relevant.3
The global incidence of NMSC continues to rise, driven by increased life expectancy, cumulative ultraviolet (UV) exposure, and improved diagnostic awareness.4,5 Demographic and environmental factors – including age, sex, and anatomical location – are known to influence NMSC development.6,7 However, the extent to which these variables independently help distinguish SCC from BCC remains incompletely characterized, particularly in region-specific populations.
Anatomically, BCC predominates on sun-exposed facial regions such as the nose and periorbital area, whereas SCC is more common on the lip, scalp, and ears. Nevertheless, population-specific analyses are essential because geographic and behavioral differences may alter these distributions.8
Most previous studies have focused on descriptive epidemiology, with limited integration of multivariate approaches to identify independent predictors. Furthermore, the potential non-linear relationship between age and tumor type has not been adequately explored.
The aim of this study was to evaluate the clinicodemographic characteristics and anatomical distribution of BCC and SCC in a Turkish cohort and to identify independent predictors of SCC using univariate and multivariate logistic regression, with particular attention to non-linear age effects.
MATERIAL AND METHODS
This retrospective study was conducted at a tertiary care center and included patients with histopathologically confirmed BCC or SCC between January 2017 and December 2024. Patients with incomplete data or rare skin malignancies (e.g., Merkel cell carcinoma, angiosarcoma) were excluded. Data collected included age, sex, tumor type, and anatomical localization. Localizations were categorized into anatomical regions. High-risk anatomical localization was predefined as lesions located on the lip, scalp, or temporal region based on their predominance among SCC cases in the present cohort and their clinical relevance in previous literatüre.9,10
Continuous variables were expressed as mean ± standard deviation and compared using independent t-tests. Categorical variables were analyzed using Chi-square tests.
Univariate logistic regression analyses were performed for each variable individually. Multivariate logistic regression was then used to identify independent predictors of SCC (dependent variable: SCC = 1, BCC = 0). Independent variables included age (analyzed both continuously and categorically: <60, 60–80, ≥80 years, with <60 as reference) to explore potential non-linear associations and enhance model robustness, sex (male = 1, female = 0), and high-risk localization (yes = 1, no = 0).
Model performance was assessed using the area under the receiver operating characteristic curve (AUC). The optimal probability cutoff for SCC classification was determined using Youden’s index (J = sensitivity + specificity − 1). Multicollinearity was evaluated using the variance inflation factor (VIF), with VIF < 5 considered acceptable. A p < 0.05 was considered statistically significant.
All statistical analyses were performed using Python (Statsmodels, Scikit-learn). This study was approved by the Institutional Ethics Committee of Istanbul Prof. Dr. Cemil Taşçıoğlu City Hospital (Approval number: 100, dated March 30, 2026). Owing to the retrospective design, informed consent was waived.
RESULTS
Demographic characteristics
A total of 374 patients were included: 209 (55.9%) with BCC and 165 (44.1%) with SCC. Mean age was 68.2 ± 12.8 years in the BCC group and 70.5 ± 12.7 years in the SCC group (p = 0.082). SCC was significantly more common in males (72.1%) than in females, whereas BCC showed a more balanced distribution (56.0% male; p = 0.003). Demographic data are summarized in Table 1.
| Parameter | BCC (n=209) | SCC (n=165) | p-value |
|---|---|---|---|
| Age (years, mean±SD) | 68.2±12.8 | 70.5±12.7 | 0.082* |
| Gender: Male, n (%) | 117 (56.0) | 119 (72.1) | 0.003† |
Anatomical distribution
Tumor localization differed significantly between BCC and SCC (p = 0.0013). BCC was most frequently located on the nose (31.6%), whereas SCC most commonly involved the lip (15.2%), scalp (13.9%), and temporal region (12.1%). Notably, no SCC cases were observed in the infraorbital or medial canthus regions. The observation that all lip lesions in our cohort were SCC is noteworthy; however, this finding should be interpreted cautiously given the single-center design and potential selection bias. Detailed data are presented in Table 2.
| Localization | BCC (n=209) (%) | SCC (n=165) (%) |
|---|---|---|
| Nose | 66 (31.6) | 23 (13.9) |
| Infraorbital | 22 (10.5) | 0 |
| Scalp | 19 (9.1) | 23 (13.9) |
| Malar | 19 (9.1) | 18 (10.9) |
| Temporal | 16 (7.7) | 20 (12.1) |
| Medial canthus | 16 (7.7) | 0 |
| Lip | 1 (0.5) | 25 (15.2) |
Chi-square test for distribution difference:p=0.0013. BCC: Basal cell carcinoma, SCC: Squamous cell carcinoma
Univariate logistic regression
Univariate logistic regression analyses [Table 3] showed that male sex (odds ratio [OR]: 1.95, 95% confidence interval [CI]: 1.25–3.05, p = 0.003) and high-risk localization (OR: 3.58, 95% CI: 2.21–5.81, p < 0.001) were significantly associated with SCC. Age was not significant (OR: 1.01, 95% CI: 0.99–1.03, p = 0.083).
| Variable | OR | 95% CI | p-value |
|---|---|---|---|
| Age (per year) | 1.01 | 0.99–1.03 | 0.083 |
| Male sex | 1.95 | 1.25–3.05 | 0.003 |
| High-risk localization* | 3.58 | 2.21–5.81 | <0.001 |
Multivariate logistic regression (continuous age)
When age was entered as a continuous variable [Table 4], male sex (OR: 1.72, 95% CI: 1.08–2.76, p = 0.024) and high-risk localization (OR: 3.22, 95% CI: 1.96–5.28, p < 0.001) remained independent predictors. Age did not reach statistical significance (p = 0.059).
| Model/variable | OR | 95% CI | p-value |
|---|---|---|---|
| Continuous age model | |||
| Male sex | 1.72 | 1.08–2.76 | 0.024 |
| High-risk localization* | 3.22 | 1.96–5.28 | <0.001 |
| Age (per year) | 1.02 | 1.00–1.04 | 0.059 |
| Categorical age model (ref: <60 years) | |||
| Male sex | 1.87 | 1.15–3.03 | 0.012 |
| High-risk localization* | 3.30 | 2.01–5.42 | <0.001 |
| Age 60–80 years | 1.31 | 0.74–2.34 | 0.352 |
| Age ≥80 years | 2.54 | 1.27–5.05 | 0.008 |
Multivariate logistic regression (categorical age)
To explore potential non-linear age effects, age was categorized with <60 years as the reference group [Table 4]. Male sex (OR: 1.87, 95% CI: 1.15–3.03, p = 0.012) and high-risk localization (OR: 3.30, 95% CI: 2.01–5.42, p < 0.001) remained significant. The 60–80 age group showed no significant increase in SCC risk (OR: 1.31, p = 0.352). In contrast, patients aged ≥80 years had a significantly higher risk of SCC (OR: 2.54, 95% CI: 1.27–5.05, P = 0.008).
The consistency of effect estimates across univariate and multivariate models supports the robustness of these findings. No multicollinearity was detected (VIF < 1.1 for all predictors).
Model performance and probability cutoff
Receiver operating characteristic (ROC) analysis demonstrated moderate discriminative ability (AUC = 0.660) [Figure 1]. The Youden’s index identified an optimal probability cutoff of 0.608 for distinguishing SCC from BCC. At this cutoff, sensitivity was 41.3% and specificity was 89.7%. The high specificity suggests that the model may be more useful for supporting the identification of SCC in high-risk cases rather than for screening purposes. This cutoff may assist in risk stratification but should not be used as a standalone diagnostic tool. Model performance metrics are summarized in Table 5.
| Metric | Value |
|---|---|
| AUC (ROC) | 0.660 |
| Optimal probability cutoff (Youden) | 0.608 |
| Sensitivity at cutoff | 41.3% |
| Specificity at cutoff | 89.7% |
| LLR p-value (Likelihood ratio test p-value) | <0.001 |
| Pseudo R2 | 0.080 |
| VIF (Age) | 1.02 |
| VIF (Sex) | 1.06 |
| VIF (High-risk localization) | 1.04 |
AUC: Area under the ROC curve, ROC: Receiver operating characteristic, VIF: Variance inflation factor, p< 0.05 was considered statistically significant
DISCUSSION
This retrospective study of 374 patients with BCC or SCC provides several clinically relevant findings. First, male sex and high-risk anatomical localization (lip, scalp, and temporal region) are strong and consistent independent predictors of SCC. The consistency between univariate and multivariate analyses suggests minimal confounding and supports the internal validity of the model. Second, the relationship between age and tumor type is non-linear: only age ≥80 years significantly increases SCC risk, whereas age 60–80 years does not. This indicates that risk becomes clinically relevant primarily in advanced age rather than increasing progressively. Third, we propose a probability cutoff of 0.608 for SCC prediction, intended as an adjunct for risk stratification rather than a definitive diagnostic tool.
The male predominance in SCC (72.1% vs. 56.0% in BCC) is consistent with existing literature and likely reflects higher cumulative UV exposure from occupational and behavioral factors.11,12 Importantly, the association remained significant after adjusting for age and localization, with stable OR across univariate (1.95), multivariate continuous-age (1.72), and multivariate categorical-age (1.87) models.
High-risk anatomical localization was the strongest predictor of SCC (OR > 3.0 in all models). This finding aligns with known UV exposure patterns and anatomical vulnerability.13 Together with the high SCC frequency on the scalp and temporal region, the observation regarding lip lesions supports a simple clinical rule: Lesions on the lip, scalp, or temporal region – especially in male patients – should raise suspicion for SCC.
Although the ear is recognized as a high-risk site for cutaneous SCC in international guidelines and previous studies, ear lesions were similarly distributed between BCC and SCC in our cohort (13 vs. 11 cases). In an additional sensitivity analysis, inclusion of the ear within the high-risk anatomical group did not improve model performance and resulted in lower discriminative ability. Therefore, the original predefined high-risk definition (lip, scalp, and temporal region) was retained. These findings may reflect cohort-specific characteristics and should be interpreted cautiously.
The non-linear age effect is a key observation. Current guidelines often consider older age a general risk factor for SCC, but our results suggest that the risk increase is not uniform.14 Patients aged 60–80 years did not have a significantly higher SCC risk than those under 60 years, whereas those over 80 years had more than double the odds. This challenges the assumption of a linear age–risk relationship and suggests that the sharp rise in SCC incidence occurs predominantly after the eighth decade, with BCC remaining the dominant NMSC subtype in the 60–80 age group.
Clinical and practical ımplications
In practical terms, the combination of male sex and high-risk anatomical localization may serve as a simple heuristic to prioritize biopsy decisions in routine clinical practice. These findings may also assist in prioritizing patients for earlier dermatological evaluation in resource-limited settings. From a clinical perspective, lesions located in high-risk regions (lip, scalp, and temporal) should raise a higher index of suspicion for SCC, particularly in elderly male patients. Although the model demonstrates only moderate discriminative ability, it provides clinically interpretable risk patterns that may complement, rather than replace, clinical judgment. Future studies incorporating histopathological and environmental variables may further improve predictive performance and enable more precise risk modeling.
The moderate AUC (0.660) indicates that while our model captures meaningful prognostic information, it does not provide highly accurate individual prediction. The high specificity (89.7%) suggests that when the model predicts SCC, it is likely correct. However, the low sensitivity (41.3%) indicates that many SCC cases would be missed if this cutoff were used alone. At the proposed cutoff of 0.608, 97 of 165 SCC cases (58.8%) would be classified as false negatives. Therefore, the proposed probability cutoff (0.608) should be viewed as a risk-stratification aid – for example, prompting a lower threshold for biopsy in patients with high-risk features – rather than a definitive diagnostic test. This cutoff has not been externally validated and should not be used in isolation to guide biopsy decisions. Prospective validation in an independent cohort is required before any clinical implementation. Additional factors such as tumor thickness, perineural invasion, immunosuppression, and cumulative UV exposure would likely improve predictive accuracy.15,16 The McFadden pseudo R2 of 0.080 indicates modest explanatory power, which is not unexpected given that the model was intentionally restricted to three readily available clinicodemographic predictors.
Strengths
Strengths include the large sample size, the consistent effect estimates across multiple modeling strategies (univariate, multivariate continuous, multivariate categorical), the formal assessment of multicollinearity VIF, the transparent reporting of model performance, and the proposal of a clinically usable probability cutoff.
Limitations
First, the retrospective, single-center design and absence of external validation may limit the generalizability and applicability of the predictive model. Second, important clinical variables – including histopathological subtype (e.g., infiltrative BCC), tumor thickness, perineural invasion, immunosuppression status, and detailed sun exposure history – were unavailable. Third, the moderate AUC underscores the need for additional predictors. Fourth, applying the proposed cutoff of 0.608 to our cohort would result in 97 false negatives (58.8% of SCC cases).
CONCLUSION
Male sex and high-risk anatomical localization (lip, scalp, and temporal region) are strong and consistent independent predictors of SCC. Age exhibits a non-linear association: SCC risk is significantly increased only in patients older than 80 years, not in those aged 60–80 years. The proposed probability cutoff of 0.608 (sensitivity: 41.3%, specificity: 89.7%) may assist in clinical risk stratification, although the moderate AUC indicates that additional variables are needed for precise individual prediction. These findings provide a simple and clinically applicable framework for identifying patients at higher risk of SCC and may support earlier diagnostic decision-making in routine clinical practice .
Authors’ contributions:
Fatih İçbudak: Conceptualization, methodology, formal analysis, statistical analysis, writing – original draft, supervision. Melih Kaan Saridal: Data collection, review and approval of the final manuscript.
Ethical approval:
The research/study was approved by the Institutional Review Board at the Institutional Ethics Committee of Istanbul Prof. Dr. Cemil Taşçıoğlu City Hospital, number 100, dated March 30, 2026.
Declaration of patient consent:
Patient’s consent is not required as patients identity is not disclosed or compromised.
Conflicts of interest:
There are no conflicts of interest.
Use of artificial intelligence (AI)-assisted technology for manuscript preparation:
The authors confirm that artificial intelligence (AI)-assisted technology was used ie. ChatGPT (OpenAI) was solely for English language editing and improving the clarity of the manuscript. All scientific content, data analysis, interpretation, and final manuscript approval were performed by the authors.
Financial support and sponsorship: Nil.
References
- Epidemiological trends, clinical impact, and geographical variations of nonmelanoma skin cancers: A twelve-year study in Messina, Italy. J Skin Cancer. 2025;2025:1823281.
- [CrossRef] [PubMed] [Google Scholar]
- The global burden of skin cancer: A longitudinal analysis from the Global Burden of Disease Study, 1990-2017. JAAD Int. 2021;2:98-108.
- [CrossRef] [Google Scholar]
- Updates on the systemic treatment of advanced non-melanoma skin cancer. Front Med. 2019;6:160.
- [CrossRef] [PubMed] [Google Scholar]
- Epidemiology of skin cancer In: Sunlight. Vitamin D and Skin Cancer. New York, NY: Springer; 2014. p. :120-40.
- [CrossRef] [PubMed] [Google Scholar]
- Epidemiological trends in skin cancer. Dermatol Pract Concept. 2017;7:1-6.
- [CrossRef] [PubMed] [Google Scholar]
- Cutaneous squamous cell carcinoma: Incidence, risk factors, diagnosis, and staging. J Am Acad Dermatol. 2018;78:237-47.
- [CrossRef] [Google Scholar]
- Risk factors and diagnosis of advanced cutaneous squamous cell carcinoma. Dermatol Pract Concept. 2021;11(Suppl 2):e2021166S.
- [CrossRef] [PubMed] [Google Scholar]
- Epidemiological and localization characteristics of non-melanoma skin cancers: Retrospective analysis of 400 cases. Turk J Plast Surg. 2019;27:19.
- [CrossRef] [Google Scholar]
- The scalp is a high-risk site for cutaneous squamous cell carcinoma metastasis. J Am Acad Dermatol. 2021;84:1742-4.
- [CrossRef] [PubMed] [Google Scholar]
- Detailed head localization and incidence of skin cancers. Sci Rep. 2021;11:12391.
- [CrossRef] [PubMed] [Google Scholar]
- Sex disparity for patients with cutaneous squamous cell carcinoma of the head and neck: A systematic review. Cancers (Basel). 2022;14:5830.
- [CrossRef] [PubMed] [Google Scholar]
- Occupational ultraviolet light exposure increases the risk for non-melanoma skin cancer: Systematic review and meta-analysis. Majmaah J Health Sci. 2022;10:46-58.
- [CrossRef] [Google Scholar]
- Anatomical exposure patterns of skin to sunlight: Relative contributions of direct, diffuse and reflected ultraviolet radiation. Br J Dermatol. 2012;167:383-90.
- [CrossRef] [PubMed] [Google Scholar]
- Epidemiology of keratinocyte skin cancer with a focus on cutaneous squamous cell carcinoma. Cancers (Basel). 2024;16:606.
- [CrossRef] [PubMed] [Google Scholar]
- Risk factors for cutaneous squamous cell carcinoma recurrence, metastasis, and disease-specific death: A systematic review and meta-analysis. JAMA Dermatol. 2016;152:419-28.
- [CrossRef] [PubMed] [Google Scholar]