Identifying risk factors for the prognosis of head and neck cutaneous squamous cell carcinoma: A systematic review and meta-analysis

In this study, we sought to identify the potential impacts of disease characteristics on the prognosis of cutaneous squamous cell carcinoma (cSCC). We searched the PubMed, EmBase, and Cochrane Library databases from their inception until February 2020 to identify studies that investigated the prognosis of cSCC. The pooled effect estimates were applied using odds ratios (OR) and 95% confidence intervals (CI) and were calculated using the random-effects model. Forty-three studies including a total of 21,530 patients and reporting 28,627 cases of cSCC were selected for the final meta-analysis. Poor differentiation (OR, 3.54; 95% CI, 2.30–5.46; P < 0.001), perineural invasion (OR, 3.27; 95% CI, 1.60–6.67; P = 0.001), Breslow greater than 2 mm (OR, 5.47; 95% CI, 2.63–11.37; P < 0.001), diameter greater than 20 mm (OR, 4.62; 95% CI, 2.95–7.23; P < 0.001), and location on temple (OR, 3.20; 95% CI, 1.12–9.15; P = 0.030) were associated with an increased risk of recurrence, whereas immunosuppression status and location on cheek, ear, or lip were not associated with the risk of recurrence. Poor differentiation (OR, 6.82; 95% CI, 4.66–9.99; P < 0.001); perineural invasion (OR, 7.15; 95% CI, 4.73–10.83; P < 0.001); Breslow greater than 2 mm (OR, 6.11; 95% CI, 4.05–9.21; P < 0.001); diameter greater than 20 mm (OR, 5.01; 95% CI, 2.56–9.80; P < 0.001); and location on ear (OR, 2.38; 95% CI, 1.39–4.09; P = 0.002), lip (OR, 2.15; 95% CI, 1.26–3.68; P = 0.005), and temple (OR, 2.77; 95% CI, 1.20–6.40; P = 0.017) were associated with an increased risk of metastasis, whereas immunosuppression status and location on cheek did not affect the risk of metastasis. Finally, poor differentiation (OR, 5.97; 95% CI, 1.82–19.62; P = 0.003), perineural invasion (OR, 6.64; 95% CI, 3.63–12.12; P < 0.001), and Breslow greater than 2 mm (OR, 3.42; 95% CI, 1.76–6.66; P < 0.001) were associated with an increased risk of disease-specific death, whereas diameter; immunosuppression status; and location on ear, lip, and temple did not affect the risk of disease-specific death. We found that differentiation, perineural invasion, depth, diameter, and location could affect the prognosis of cSCC. The potential role of other patient characteristics on the prognosis of cSCC should be identified in further large-scale prospective studies.


Introduction
Nonmelanoma skin cancer is the most common type of cancer, and the number of cases has increased to more than 1 million in the United States annually [1][2][3]. Cutaneous squamous cell carcinoma (cSCC) accounts for 20%-30% of nonmelanoma skin cancer, and most patients present with a head or neck location [4]. Most cases of cSCC can now be cured through surgery, but a certain subset of patients with high-risk cSCC could experience local recurrence, nodal metastasis, and disease-specific death (DSD) [5]. Although the prevalence of local recurrence, nodal metastasis, and DSD was low, the prognosis of these cases was poor, and the 5-year survival rate was only 30% to 60% for patients treated surgically for lymph node metastases from cSCC of head and neck [6].
The incidence of head and neck cSCC is increasing, and the specific anatomic subsites have been associated with an increased risk of regional metastatic involvement [7]. The parotid gland has been identified as a common location for spread because it could receive lymphatics from the ear, forehead, face, temple, and scalp [8]. Cancers with involvement of the parotid gland frequently demonstrated macroscopic or microscopic cervical metastases [9]. Other clinicopathological features, including differentiation, perineural invasion, depth, diameter, and immunosuppression, might also affect the prognosis of cSCC [10]. However, the predictive value of these clinicopathological features on the prognosis of cSCC vary. Therefore, the current systematic review and metaanalysis was conducted based on all available evidence to obtain pooled results regarding the predictive value of clinicopathological characteristics on the prognosis of cSCC.

Data sources, search strategy, and selection criteria
This study was performed and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Statement [11]. Studies with an observational design that investigated the role of clinicopathological characteristics on the prognosis of cSCC were eligible for inclusion in this study, and restrictions were not applied for publication language or status (published, in press, or in progress). The electronic databases of PubMed, EmBase, and the Cochrane Library were systematically searched from their inception until February 2020 using the following search terms: (skin or cutaneous or dermal or cutanea) AND (squamous or epidermoid or planocellular or "prickle cell" or verrucous) AND (carcinoma � ) AND (outcome � or recurrence � or relaps � or recrudescence � or recurrent or recidive or metastas � or metastatic � or spread � or disseminat � or secondary or migrat � or death or morbidity or mortality or surviv � ) AND (risk or risks or "perineural invasion � " or "peri-neural invasion � " or PNI or depth or thickness or size or diameter or or location � or ear or ears or cheek � or lip or lips or differentiation or immunocompromised or "immune compromi � " or "sentinel lymph node � "). Details regarding the search strategy are presented in S1 Appendix. The reference lists of retrieved studies were also reviewed manually to identify any additional studies that met the inclusion criteria.
Two reviewers independently conducted the literature search and study selection. Any disagreements were resolved by discussion until a consensus was reached. A study was included if it met all of inclusion criteria: (1) all patients had cSCC; (2) the study reported at least 1 of the following: differentiation, perineural invasion, depth, diameter, immunosuppression status, and location; and (3) the study reported at least 1 of the following: recurrence, metastasis, and DSD. Furthermore, there were no restrictions placed on study design.

Data collection and quality assessment
The information of included studies were collected by 2 reviewers independently. The abstracted items included the first author's name, publication year, study design, country, sample size, mean age, number of cSCCs, outcomes, risk factors, inclusion criteria, duration of follow-up, and investigated outcomes. The quality of included studies was assessed using the Newcastle-Ottawa Scale (NOS) according to selection (4 items, 4 stars), comparability (1 item, 2 stars), and outcome (3 items, 3 stars), and the "star system" for each individual study ranged from 0 to 9 [12]. The quality assessment was conducted by 2 reviewers, and any inconsistencies were resolved by an additional reviewer who referred to the full-text of the article.

Statistical analysis
The effect estimates for the role of clinicopathological characteristics on the prognosis of cSCC were assigned as odds ratios (OR) and 95% confidence intervals (CI), because the analysis included both prospective and retrospective studies. The random-effects model was applied to calculate the pooled effect estimates because it could consider the underlying variations across the included studies [13,14]. I 2 and the Q statistic were applied to assess the heterogeneity among the included studies. Significant heterogeneity was defined as I 2 greater than 50.0% or a P value less than 0.10 [15,16]. Sensitivity analysis was conducted for outcomes that reported 5 or more studies to assess the robustness of the pooled conclusions by sequential exclusion of individual studies [17]. Publication bias for outcomes that reported 5 or more studies was assessed using funnel plots, and the Egger and Begg tests [18,19]. P values of less than 0.10 were considered as potential publication bias for the Egger and Begg tests. The inspection level for pooled results was 2-sided, and a P value less than 0.05 was regarded as statistically significant. All statistical analyses in this study were conducted using the STATA software (version 10.0; StataCorp, College Station, TX).

Literature search
The electronic searches in PubMed, EmBase, and the Cochrane Library yielded 5946 articles, 2794 of which were retained after duplicate records were excluded. Of the remaining articles, 2711 were excluded because they reported irrelevant titles. The remaining 83 studies were retrieved for further full-text evaluation. Forty more studies were excluded for the following reasons: insufficient data (n = 18), other disease status (n = 13), and review or meta-analysis (n = 9). A total of 43 studies were selected for the final meta-analysis (Fig 1)  . Nineteen relevant studies were identified through reviewing the reference lists of the included studies. However, these studies were excluded as duplicate records because they had been included in the initial electronic searches. Table 1 summarizes the baseline characteristics of the included studies and patients. A total of 21,530 patients and 28,627 cSCCs were included. Seven studies were designed as prospective, and the remaining 36 studies were retrospective. Twenty-five studies were conducted in Europe, 17 studies were conducted in America, and 1 study was conducted in Asia. The sample size ranged from 41 to 6164, and the duration of follow-up ranged from 1.0 to 10.0 years. The quality of the studies was assessed using the NOS, and the number of NOS stars ranged from 5 to 9, with the median of 8 stars (interquartile range, 7-9 stars).

Discussion
The current study was based on observational studies and explored the role of clinicopathological characteristics on the risk of recurrence, metastasis, and DSD. This comprehensive quantitative study included 21,530 patients and 28,627 cSCC from 7 prospective and 36 retrospective studies with a wide range of patient characteristics. The findings of this study found that poor differentiation, perineural invasion, and Breslow greater than 2 mm were associated with an increased risk of recurrence, metastasis, and DSD. Moreover, diameter greater than 20 mm and location on the temple were associated with excess risk of recurrence and metastasis. Finally, location on the ear or lip was associated with an increased risk of metastasis. A previous systematic review and meta-analysis included 36 studies and reported that the risk of recurrence was significantly increased with Breslow greater than 2 mm, invasion beyond subcutaneous fat, Breslow greater than 6 mm, perineural invasion, and poor differentiation. Invasion beyond subcutaneous fat; Breslow greater than 2 mm; Breslow greater than 6 mm; diameter greater than 20 mm; poor differentiation; perineural invasion; location on temple, ear, and lip; and immunosuppression were associated with an increased risk of metastasis.  Furthermore, diameter greater than 20 mm, poor differentiation, location on ear or lip, invasion beyond subcutaneous fat, and perineural invasion were associated with an excess risk of DSD [10]. However, the additional published articles should be entered into a meta-analysis to reevaluate the role of clinicopathological characteristics on the prognosis of cSCC. Therefore, we conducted this systematic review and meta-analysis to provide an accurate predictive value of clinicopathological characteristics on the prognosis of cSCC. The potential role of clinicopathological characteristics on the risk of recurrence was consistent with that reported in a previous meta-analysis [10]. Nearly all of studies reported similar results or trends. Moreover, most identified prognostic factors for the risk of metastasis were similar to the previous study. However, immunosuppression status was associated with a nonsignificant risk of metastasis, which was not consistent with the previous study [10]. The potential reasons could be accurate follow-up of immunocompromised patients for early diagnosis of cSCC patients in tertiary care centers and that the number of high-risk cSCCs among immunocompromised and immunocompetent patients were not balanced, which could affect the prognosis of cSCC. Moreover, although the role of most disease characteristics on the risk of DSD were consistent with a previous meta-analysis, the role of diameter, location on ear, and location on lip on the risk of DSD were inconsistent with the previous study. The potential reason for this could be that the results of previous study for the role of diameter, location on ear, and location on lip on the risk of DSD were based on only a single study. These results vary, and further large-scale prospective studies are needed for verification.
Several advantages of this study should be highlighted. First, most of the included studies were of relative high quality, and the pooled conclusions were robust. Second, the large sample size allowed us to quantitatively assess the role of clinicopathological characteristics on the prognosis of cSCC. Thus, our findings are potentially more robust than are those of any individual study. Third, the analysis of this study provides comprehensive results regarding the role of differentiation, perineural invasion, depth, diameter, immunosuppression status, and location on the risk of recurrence, metastasis, and DSD.
Despite these advantages, however there are also limitations that should be acknowledged. Most of the included studies were retrospective in design, and selection or recall bias was inevitable. In addition, significant heterogeneity was detected for several of the investigated outcomes, which was not fully interpreted by sensitivity analyses. Third, most of the reported results were unadjusted data, and the results could be affected by the patient characteristics. Fourth, the role of clinicopathological characteristics on the risk of DSD was reported in a smaller number of studies, and the conclusions might vary. Also, publication bias was inevitable because of the analysis based on published articles. Finally, the analysis was based on the study level, and individual data were not available.

Conclusion
In conclusion, this study indicated that the risk of recurrence, metastasis, and DSD among cSCC patients are significantly increased when patients are characterized by poor differentiation, perineural invasion, and Breslow greater than 2 mm. Moreover, we noted that diameter greater than 20 mm and location on the temple were associated with an increased risk of recurrence and metastasis. Furthermore, location on the ear or lip could result in an excess risk of metastasis in patients with cSCC. The results of this study should be verified by further prospective cohort studies.