Nonmelanoma Skin Cancer With Aggressive Subclinical Extension in Immunosuppressed Patients

Importance  Immunosuppression (IS), such as in solid-organ transplant recipients (SOTRs) and patients with human immunodeficiency virus (HIV) or hematologic malignant neoplasms, increases the risk of developing nonmelanoma skin cancers (NMSCs). However, it is unknown whether IS patients are at increased risk of developing NMSCs with aggressive subclinical extensions (NMSC-ASE), which may extend aggressively far beyond conventional surgical margins.

Objective  To study clinical characteristics of NMSC-ASE among immunocompetent (IC) and various subgroups of IS patients and to suggest a predictive model for NMSC-ASE lesions.

Design, Setting, and Participants  A 6-year retrospective review of 2998 NMSC cases between February 26, 2007, and February 17, 2012, at the Dermatologic and Mohs Micrographic Surgery Unit of the University of California, San Diego, Medical Center. Nonmelanoma skin cancers that required at least 3 Mohs micrographic surgery stages with final surgical margins of at least 10 mm were defined as ASE lesions. All cases were categorized into 1 of 2 groups, IS or IC. Immunosuppressed cases were further subcategorized into 3 subgroups: SOTRs and patients with HIV or hematologic malignant neoplasm. The data were analyzed in December 2012.

Main Outcomes and Measures  We evaluated the odds ratio of having NMSC-ASE lesions in IS patients (SOTRs, HIV, hematologic malignant neoplasm) compared with IC patients. Other clinical characteristics and preoperative risks were analyzed and compared.

Results  Of all 2998 cases, we identified 805 NMSC-ASE cases: 137 IS and 668 IC. Immunosuppressed patients had an odds ratio of 1.94 of having ASE lesions compared with IC patients (95% CI, 1.54-2.44; P < .001). Additionally, the SOTR subgroup was associated with a 2.74 odds of having NSMC-ASE compared with non-SOTRs (95% CI, 2.00-3.76; P < .001), and the presence of hematologic malignant neoplasm was associated with 1.74 times the odds compared with IC patients (95% CI, 1.04-2.90; P = .04). Multivariate analysis found older age (P < .001), lesion locations such as zone 1 (OR, 1.39 [95% CI, 1.04-1.85]; P = .02) or zone 2 (OR, 1.45 [95% CI, 1.08-1.94]; P = .01), and IS status (OR, 1.94 [95% CI, 1.54-2.44]; P < .001) to be significant predictors of ASE.

Conclusions and Relevance  The findings of this study suggest an increased risk for NMSC-ASE lesions in IS patients, especially in SOTRs and those with hematologic malignant neoplasm, but not patients with HIV. Statistically significant predictors of NMSC-ASE lesions such as age, location, and IS status can help physicians choose the most appropriate treatment modalities and optimize surgical planning.

Nonmelanoma skin cancer (NMSC) is the most common malignant neoplasm in the United States¹ and the most commonly diagnosed skin cancer worldwide in those with fair skin and increased sun exposure.² Basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) are the most common types of NMSC in the United States, with a ratio of approximately 4:1, respectively.² The clinically visible size of NMSC lesions on the skin surface does not always reflect the degree of invasion into the dermis or subclinical extension.³ Therefore, standard surgical margins are determined according to the pathologic diagnosis of the tumor, as well as the clinical presentation. For most common types of NMSCs, the current recommended excision margin is 4 to 5 mm,⁴ but these margins are usually set wider for lesions greater than 2 cm in diameter. However, local microscopic invasion can extend beyond standard margins in some cases, even for smaller clinical lesions.⁴ There is evidence in the current literature that certain high-risk histologic subtypes of NMSCs have aggressive subclinical extension (ASE), but the importance of other predisposing factors, such as immunosuppression (IS), has not been clearly defined.^5-8 Identification of such predisposing factors for ASE is important to help guide physicians in determining appropriate surgical margins to allow for complete tumor excision and subsequently reduce the risk of recurrence or metastasis.

A review of the current literature shows only 1 study that specifically investigated the predictors of extensive subclinical spread of all types of NMSCs treated with Mohs micrographic surgery (MMS). In 2002, Batra and Kelley⁵ examined possible risk factors for subclinical spread to help identify aggressive NMSCs that otherwise might be inadequately treated. Their retrospective study defined extensive tumor spread as at least 3 MMS stages without consideration of the final surgical margins. Their data showed that significant predictors of extensive subclinical spread included histologic subtype, location of lesion, recurrence, and preoperative size greater than 25 mm. Contrary to previous studies, however,^2,9 their data showed that immunocompromised status was not a significant predictor of extensive subclinical invasion. Our study attempts to clarify these conflicting clinical observations by collecting a greater number of cases, stratifying the data according to cause of immunocompromised status, and using more stringent criteria for ASE (MMS ≥3 stages and final surgical margins ≥10 mm), as defined by our previous study on SCC with ASE,¹⁰ to account for any confounding nonaggressive lesions in cosmetically sensitive areas.

Box Section Ref ID

In this study, we conducted a 6-year comparison of clinical data between IS and immunocompetent (IC) patients with the purpose of elucidating whether immune status is a reliable risk factor for NSMC-ASE. In addition, we stratified IS patients into 3 groups to better identify differences in the clinical characteristics of NMSC-ASEs based on the etiologic mechanism of immune suppression: solid-organ transplant recipients (SOTRs) and patients with human immunodeficiency virus (HIV) and hematologic malignant neoplasms. Last, we present a predictive model using significant clinical characteristics from our cohort suggestive of NSMC-ASE lesions.

A retrospective review of patient data was performed at the Dermatologic and Mohs Micrographic Surgery Unit of the University of California, San Diego (UCSD), Medical Center. Male and female patients 18 to 100 years of age with either IC or IS state and NMSCs were identified using the UCSD electronic medical record. All MMS cases from this cohort with a biopsy-confirmed diagnosis of BCC and SCC presenting between February 26, 2007, and February 17, 2012, were included in this study. The study protocol was approved by the institutional review board of UCSD prior to data collection. Informed consent was waived due to the retrospective nature of the study. All data were gathered via an electronic medical record with secondary confirmation from paper records of all MMS cases for those procedures occurring before 2008. Personal identifiers were removed, and data were encrypted for confidentiality.

Patient characteristics such as age at surgery, sex, and immunological status were collected. Clinical characteristics included location of tumor, preoperative and postoperative dimensions, final surgical margins (defined as the difference between postoperative and preoperative defect dimensions at both x and y axis), number of stages required to achieve histologically clear margins, and histologic diagnosis. The tumor sizes were recorded as 2 orthogonal dimensions (x and y axis) by the surgeon as preoperative size I and II, and postoperative size I and II. Smaller orthogonal length of the tumor was recorded as size I, while longer length was recorded as size II. Consequently, differences between preoperative and postoperative size I and II, respectively, yielded 2 final surgical margin values for each tumor.

Aggressive subclinical extension was defined as at least 3 MMS stages and final surgical margin at least 10 mm in at least 1 dimension as previously reported by our group.¹¹ The final surgical margin of at least 10 mm was chosen because each MMS stage is 2 to 3 mm in size, and 3 stages will yield at least a 6 to 9 mm final margin. This additional requirement will prevent lesions in cosmetically sensitive areas, where smaller MMS stages are taken, to be inaccurately included in the data analysis. Non-ASE and NMSC-ASE cases were arranged into 2 statistically independent groups so as to avoid double inclusion of the same participants within both groups. All patients were categorized into 1 of 2 groups, IS or IC. Immunosuppressed cases were further divided into 1 of 3 subgroups (HIV, SOTR, or hematologic malignant neoplasm). The SOTR group was further classified as heart, kidney, liver, or lung only. Double transplantation cases and those with multiple causes of IS (eg, SOTR and hematologic malignant neoplasm) were excluded from our analysis.

Lesion location was defined in 2 distinct ways: first by anatomical location, and second by zones consistent with the 2013 National Comprehensive Cancer Network Guidelines¹¹ in order to strengthen the power of each area. As such, zone 1 was defined as the “mask areas” of the face (central face, eyelids, eyebrows, periorbital, nose, lips (cutaneous and vermillion), chin, mandible, preauricular and postauricular skin/sulci, temple, ear), genitalia, hands, and feet. Zone 2 consisted of the cheeks, forehead, scalp, neck, and pretibial area; zone 3 was defined as the trunk and extremities excluding pretibial, hands, feet, nail units, and ankles. These anatomical areas also correspond to the H, M, and L areas, respectively, as delineated in the American Academy of Dermatology’s appropriate use criteria on MMS.¹²

Cases of BCC were subdivided into 4 distinct groups with similar presumed aggression levels based on histologic features described in prior studies¹³ and surgeon experience to increase the power to detect differences among higher sampled groups. Group 1 included superficial and superficial multicentric BCC; group 2 included cystic, nodular, pigmented, ulcerated, unspecified, and adenoid BCC; group 3 included basosquamous, metatypical, micronodular, infiltrative, morpheaform, and sclerosing BCC; group 4 was solely recurrent BCC. Cases of SCC were also subdivided into 4 distinct groups: group 1 included SCC in situ and Bowen; group 2 included “not otherwise specified” lesions; group 3 was poorly and moderately differentiated; group 4 was solely recurrent SCCs.

Data were analyzed via an independent t test for continuous variables and χ² or Fisher exact test for binary variables. A logistic regression analysis using a backward likelihood ratio technique with removal set at P = .10 was performed using SPSS, version 21 (IBM Corp, 2012). The primary outcome of interest in our regression model was the development of ASE, and the covariates tested were histologic features, age, sex, skin type, location of lesion, and patient IS status.

Between February 2007 and February 2012, 4037 MMS procedures were performed at the UCSD Dermatologic and MMS unit; 2998 NMSCs (2044 BCCs and 954 SCCs) were included in the analysis. Of these, 2193 (73%) cases had no history of concurrent or prior NMSC-ASE, and 805 (27%) cases were of NMSC-ASE (Table 1).

Mean (SD) age was significantly higher in the NMSC-ASE group (68.3 [14.1] vs 65.8 [14.5] y; P < .001). The proportion of males to females was not significantly different between the NMSC-ASE and non-ASE groups (P = .13). The majority of cases in both the NMSC and NMSC-ASE groups were within individuals of Fitzpatrick skin type II (61% and 60%, respectively). There was no significant difference in Fitzpatrick skin type between the 2 groups. Tumor location by zone showed that tumors in zone 1 (OR, 1.39 [95% CI, 1.04-1.85]; P = .02) and zone 2 (OR, 1.45 [95% CI, 1.08-1.94]; P = .01) were more likely to be NMSC-ASE than those of the less aggressive zone 3. When location was analyzed by anatomical site, the eyebrow (OR, 2.57 [95% CI, 1.64-4.03]; P < .001), lip (OR, 1.75 [95% CI, 1.04-2.94]; P = .03), and nose (OR, 1.99 [95% CI, 1.09-3.64]; P = .03) were significantly associated with NMSC-ASE when compared with the trunk (chest and back). The pretibial area was associated with a decreased odds of NMSC-ASE (OR, 0.38 [95% CI, 0.16-0.89]; P = .02) compared with the trunk.

Overall, BCCs had 1.40 times the odds of being subclinically aggressive compared with SCCs (95% CI, 1.19-1.67; P < .001). The group 2 BCC subtype (cystic, nodular, pigmented, ulcerated, unspecified, adenoid) was predominant in both BCC and BCC-ASE groups (79% and 71%, respectively.) Basal cell carcinomas falling into subtype group 3 (basosquamous, metatypical, micronodular, infiltrative, morpheaform, sclerosing) were significantly more likely to be BCC-ASE (OR, 1.65 [95% CI, 1.26-2.17]; P < .001). The majority of SCCs were within groups 1 and 2 (in situ and not otherwise specified, respectively.) Cases of SCC with a not otherwise specified distinction were less likely (OR, 0.69 [95% CI, 0.52-0.92]) to have ASE when compared with in situ SCCs (P = .01). Preoperative sizes of NMSC-ASEs were significantly larger than those of non-ASE NMSCs (P < .001).

Overall, IS patients had 1.94 times the odds of having NMSC-ASE compared with IC patients (95% CI, 1.54-2.44; P < .001). When further subdivided by IS type, SOTRs were 2.74 times more likely to have NMSC-ASE than non-SOTRs (IC, HIV, hematologic malignant neoplasm group). Subgroup analysis of SOTRs found that patients with heart and lung transplants were significantly more likely to have NMSC-ASE than those with kidney transplants (OR, 3.22 [95% CI, 0.96-10.88]; P = .06; and OR, 14.50 [95% CI, 1.80-116.36]; P = .01, respectively). Patients with hematologic malignant neoplasms had 1.74 times the odds of having NMSC-ASE compared with IC patients (95% CI, 1.04-2.90; P = .04).

Results were further adjusted using a logistic regression model that organized location by zones (Table 2). Age, SOTR status, history of hematologic malignant neoplasm, and location in zones 1 and 2 remained significant predictors for NMSC-ASE, consistent with univariate findings in Table 1. Nonmelanoma skin cancers within zones 1 and 2 were found to be significantly associated with ASE both in univariable and multivariable analyses (P = .04, .02, respectively). Solid-organ transplant recipients and patients with hematologic malignant neoplasm were significantly more likely to have NMSC-ASE than IC participants when results were controlled for demographic factors such as age, sex, and NMSC type (OR, 3.09 [95% CI, 2.23-4.27]; P < .001; OR, 1.90 [95% CI, 1.12-3.20]; P = .02, respectively).

Based on our logistic regression model, the probability of NMSC-ASE was determined using the predictive capacity of the chosen variables from information on the covariates (age, IS status, location) (Table 3). Because sex was not a significant predictor of ASE, it was not included in our probability prediction model. The Figure is a visual representation of Table 3 results, depicting the percentage probability of a patient having NMSC-ASE (if a diagnosis of NMSC is received) based on their clinical characteristics, age, and location of the lesion—all factors that can be gathered prior to initiation of MMS. The percentage probability of NMSC-ASE is highest for older patients regardless of any of the clinical characteristics. Among the differing clinical risk factors, SOTRs have the highest percentage probability of NMSC. Patients with lesions in zones 1 and 2 have overall higher percentage probability of having NMSC-ASE if they receive a diagnosis of NMSC than those with lesions within zone 3. As an example, based on our model, a 70-year-old SOTR with a superficial BCC on the nose (zone 1) has a 50% probability of having NMSC-ASE if a diagnosis of NMSC is received. In comparison, an IC patient of the same age and lesion presentation has a 24% chance of having NMSC-ASE when receiving a diagnosis of NMSC.

Previous studies have shown that IS patients, including SOTRs and patients with HIV or hematologic malignant neoplasm, have an increased risk of developing multiple, aggressive NMSCs.⁹ There have also been reports of IS patients having increased risk of metastasis from NMSCs.¹⁴ Our study reaffirms these findings that IS not only increases the frequency of NMSCs,¹⁵ but also that among IS patients, SOTRs and those with hematologic malignant neoplasm are at especially increased odds of NSMC-ASE when compared with other IS patients.^16,17

When further subdivided into solid-organ transplant type, those with lung and heart transplantation were more likely to have NMSCs with ASE (OR, 14.50, P = .01; OR, 3.22, P = .06, respectively). Prevalence of skin cancer in lung transplant recipients at 1, 5, and 10 years after transplantation has been reported to be 0.9%, 8.2%, and 24.7%, respectively.¹⁸ It is likely that lung and heart transplantation are associated with a higher likelihood of ASE lesions because these patients tend to receive more overall immune suppression than other SOTRs such as kidney or liver transplant recipients.¹⁹ Immunosuppressants not only compromise immune surveillance but also increase skin cancer risk by substance-specific mechanisms: blocking cellular DNA repair mechanisms in UV-damaged cells,²⁰ causing oxidative DNA damage,²¹ and upregulating p53-oncogene.²² Of all the subtypes of SOTR, lung transplant recipients had the highest association with NMSC-ASE lesions, likely because of their more potent IS regimen.²³ In addition to their IS regimen, use of long-term prophylactic antibiotics and antifungal medications given to lung transplant recipients, such as voriconazole, has been associated with an increased risk of SCC in recently published articles.^24,25 One possible mechanism of action may be the induction of UV-mediated DNA damage or disruption of DNA damage repair. However, these observations are from relatively small case series, and a subsequent larger study of lung transplant recipients showed no association between aggressive skin cancer and voriconazole use.²³

The presence of hematologic malignant neoplasm was also found to be a significant IS subtype with increased NSMC-ASE (OR, 1.74; P = .04). The mechanism for this association is similar to that for SOTRs in that potent IS is likely an important contributing factor.²⁶ The chemotherapeutic agents used in non-Hodgkin lymphoma, for example, have severe IS effects that can increase the risk for aggressive skin cancers.^16,27 However, some studies also suggest that increased susceptibility to these aggressive cancers may be due to the leukemic process and not just the effects of treatment.²⁸ For example, dysfunctional B cells play an important role in IS in non-Hodgkin lymphoma,²⁹ while inability of lymphocytes to elicit antitumor responses in patients with chronic lymphocytic leukemia may increase the risk for secondary cancers.²⁸ As a result, it is likely that multiple mechanisms for IS in patients with hematologic malignant neoplasms make them more vulnerable to aggressive secondary cancers such as NSMC-ASE.

One notable finding in our study is that the group of patients with HIV was not a significant IS subtype with increased NSMC-ASE (OR, 1.09; P = .73). This finding helps explain the discrepancy noted in the study by Batra and Kelley,⁵ which did not show IS as a predictor of ASE. The majority of patients in their relatively small sample of 39 patients who were classified as immunocompromised had HIV. This could have influenced their conclusion that IS status was not a significant predictor of ASE lesions. Furthermore, there was no association between the degree of IS (CD4 counts) and development of ASE lesions in patients with HIV. This is consistent with the literature, which suggests that there is no correlation between the incidence of NMSC and the degree of IS in patients with HIV. CD4 counts,³⁰ viral RNA levels, and use of antiretroviral therapy in patients with HIV have not been found to be significantly associated with presence of NMSC³⁰ or increased morbidity and mortality.^30,31 Our study results may further support our theory that the potent iatrogenic IS in SOTRs or patients with hematologic malignant neoplasms may have a stronger association with ASE lesions.

Prediction of NMSC-ASE lesions based on preoperative information (age, IS status, and location) is critical not only for surgical planning but also for reducing recurrence risk by treating the lesion with sufficiently adequate margins. Several studies have attempted to create a risk scale or model for assessing preoperative risk for aggressive NMSCs.^5,32,33 However, these studies included fewer than 1500 cases, used a broad definition for aggressiveness, and did not consider the confounding effects of multiple tumors in each patient. Our study, in comparison, included nearly 3000 cases and used a more strict definition of ASE that included surgical margins of at least 1 cm. In addition, our multivariate analysis identified significant predictive factors of NMSC-ASE, namely, age, IS status, and tumor location.

One of the important advantages of our study is that we attempted to avoid comparison of each patient to themselves. In general, a complicating factor of retrospective data analyses is that a single patient may have more than 1 lesion. Because it is possible that 2 lesions from the same patient are more similar than 2 lesions from 2 different patients, not eliminating this confounder jeopardizes the independence of the groups. Prior studies have either treated each lesion independently despite some patients having multiple lesions,⁵ or decreased their case numbers by only including 1 lesion from each patient.³² Our study attempted to address this possible confounding factor by ensuring that a patient was categorized to only 1 independent group. Each patient was assigned to either non-ASE or NMSC-ASE to avoid double inclusion of the same participant within both groups.

Some limitations of this study should be noted. First, the study was conducted using a retrospective analysis of clinical data. While the data set attempted to capture all surgical cases within our Mohs clinical practice, our results were limited to data that could be derived from our electronic medical record. Second, classification of tumors into multiple subcategories based on IS location and histological diagnosis resulted in smaller sample sizes for each subtype. Some of the categories warranted combining with others to have sufficient power to draw a conclusion. Because of the small sizes of a few subcategories, this may have resulted in false-negative results for certain predictive factors of NMSC-ASE. We were also unable to classify SOTRs by the degree of IS received in our study sample. Future studies are needed to investigate the association between the potency of IS therapy and NMSC-ASE.

Reliance on visual estimates of the tumor and conventional wide surgical margins may not be useful when treating patients with NMSCs with extensive subclinical spread. Our results revealed a more accurate multivariate predictive model for ASE lesions by using the following significant variables: age, IS status, and location. The findings from our study may help dermatologists optimize the extent of excisional margins for tumors with substantial subclinical spread while allowing surgeons and patients to anticipate a larger and more extensive final surgical defect. In settings such as developing countries where MMS is not readily available, the findings from this study can help guide physicians in determining more appropriate excisional margins when treating patients with NMSCs. Further studies are needed to evaluate for associations between ASE and local and/or regional recurrence as well as metastasis, and the potential of ASE inclusion in NMSC staging criteria.

Accepted for Publication: January 28, 2016.

Corresponding Author: Shang I. Brian Jiang, MD, University of San Diego Dermatologic and Mohs Micrographic Surgery Clinic, University Pacific Center Building, 8899 University Center Ln, Ste 350, San Diego, CA 92122 (s2jiang@mail.ucsd.edu).

Published Online: March 16, 2016. doi:10.1001/jamadermatol.2016.0192.

Author Contributions: Drs Song and Jiang had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Song, Jiang.

Acquisition, analysis, or interpretation of data: Song, Goldenberg, Ortiz, Eimpunth, Oganesyan, Jiang.

Drafting of the manuscript: Song, Goldenberg.

Critical revision of the manuscript for important intellectual content: Song, Goldenberg, Ortiz, Eimpunth, Oganesyan, Jiang.

Statistical analysis: Song, Goldenberg, Jiang.

Administrative, technical, or material support: Jiang.

Study supervision: Ortiz, Jiang.

Conflict of Interest Disclosures: None reported.