Topic Article:
Chronic Diseases Requiring Hospitalization and Risk of Non-Melanoma Skin Cancers—A Population Based Study from Denmark
Annette Ø. Jensen, Anne B. Olesen, Claus Dethlefsen, Henrik T. Sørensen and Margaret R. Karagas
Journal of Investigative Dermatology (2008) 128, 926–931; doi:10.1038/sj.jid.5701094

Chronic Diseases and Non-Melanoma Skin Cancer: Is There An Association?

Yvonne Romagosa¹, Shasa Hu¹ and Robert S. Kirsner¹

Journal of Investigative Dermatology (2008) 128, 768; doi:10.1038/jid.2008.50

Non-melanoma skin cancer (NMSC) has been associated with several other conditions, including melanoma and non-Hodgkin’s lymphoma (Hjalgri et al., 2000; Hu et al., 2005). In addition, the risk of NMSC has been found to be greater in various chronic disease groups (Frisch et al., 1996; Frisch and Melbye, 1995). It is uncertain whether this association is related to the treatments for those chronic diseases (such as radiation therapy or immunosuppressant therapy) rather than the disease itself (Saladi and Persaud, 2005; Karagas et al., 2001). Moreover, using a Danish nationwide cohort of patients with NMSC, Jensen et al. (2006) found a lower mortality in basal cell carcinoma (BCC) patients compared with the general population, reflecting what the authors suggested to be a selection bias to that data set.

In a follow-up study utilizing a nationwide Danish patient registry linked to the nationwide cohort of patients with NMSC, the researchers attempted to analyze the development of NMSC in patients hospitalized for various chronic diseases (Jensen et al., 2008). To control for potential surveillance bias, they also evaluated the severity of NMSC diagnosed. Several chronic diseases were associated with increased incidence of NMSC, including an association between BCC and transplantation as well as lymphoma. The risk of squamous cell carcinoma was found to be higher in patients previously hospitalized for leukemia as well as other skin diseases.

Through the following questions, we examine this paper in greater detail.

QUESTIONS

1. Why would patients with other conditions be at risk for skin cancer?

2. What is the Charlson index, and how was it used in this study?

3. How did the investigators control for potential bias in this study?

4. What were the major findings of this study?

5. How did the investigators explain their findings?

6. What may be the clinical implications of this article?

7. What future studies would help confirm these hypothesis-generating results?

ANSWERS

1. There are at least three reasons that patients with other chronic diseases would be at risk for developing nonmelanoma skin cancer (NMSC).

First, the condition itself may allow skin cancer to “survive” more readily. For example, lymphoproliferative conditions lead to immune suppression, compromising surveillance against developing malignancies and leading to a greater risk of developing skin cancer (Hemminki et al., 2003). Likewise, conditions that cause increased physiologic stress may indirectly lead to immune suppression through increases in stress byproducts such as cortisol, which in turn increase immune suppression (Agarwal and Marshall, 1998).

Second, treatments used for chronic conditions may put patients at risk for developing NMSC. For example, treatments may cause immunosuppression, as in the case of transplant patients who receive antirejection medications (Lutzner, 1984). Additionally, some treatments may be directly carcinogenic, as seen with radiation therapy (Martin et al., 1970) and phototherapy (Hönigsmann et al., 1980).

A third reason for a patient with chronic disease to be at risk for NMSC may be a shared risk factor between the condition and skin cancer. An example of this is cigarette smoking, which is a risk factor for both chronic obstructive pulmonary disease (COPD) and NMSC (De Hertog et al., 2001). This does not mean that COPD causes skin cancer, but patients with this chronic condition may be at a higher risk for skin cancer because of the common risk from smoking.

2. Developed in 1987, the Charlson Comorbidity Index (CCI) is based on 1-year mortality data from patients admitted to a single hospital over a period of 1 month in 1984 (N = 604). The data were validated, retrospectively, with a cohort of patients with breast cancer. Overall mortality for 19 medical conditions was then converted to a relative risk (RR) of death within 1 year. This information was then weighted using a scale of 1 to 6, with 6 representing highest risk, based on the RR (Table 1). From this, a sum score gives a patient’s comorbidity score, which can also account for increasing age with the addition of 1 point for each decade of life over the age of 50 (Charlson et al., 1987).

Table 1. Example of weighted index of comorbidity

Although validated, there are several limitations to this index, including the relatively small number of subjects used to create the index, the inclusion of only 19 chronic diseases, and its limited ability to predict short-term outcomes. Based on 1984 data, the CCI gives the same weight to AIDS that it does to metastatic solid tumor. These conditions may no longer have the same level of comorbidity since the advent of antiretroviral therapy for AIDS, which at the time the index was created had an extremely poor prognosis. In this study, the investigators used the CCI to stratify study subjects into various comorbidity levels.

3. The database was created in Denmark and validated for that population. This NMSC database consisted of cancers seen by Danish dermatologists in private or public outpatient dermatology clinics in 1995 (N = 6,529), and the data were collected in a prospective fashion. Information was compiled on tumor size, location, clinical and histopathologic diagnoses, treatment, and history of previous NMSC. A second source of information for the database—patient visits to non-dermatologists, captured in examination of pathology reports (N = 4,220)—included the type of clinic, histopathologic diagnosis, and details regarding tumor growth pattern and differentiation (if available). This information was supplemented using the Danish Cancer Registry to ascertain a previous history of NMSC. The NMSC database was then linked to the Danish National Patient Registry, a database that has recorded 99.4% of all hospital discharge reports since 1 January 1977.

In an earlier Danish cohort study of NMSC patients, the investigators found a lower mortality than in the general Danish population. This led the investigators to suspect that surveillance bias may be responsible for the observed links between chronic disease and NMSC (Jensen et al., 2006) The presence of surveillance bias was evaluated by the investigators by comparing the risk of NMSC across comorbidity levels of the CCI. The investigators also studied the risk of highly aggressive tumor behavior vs. less aggressive tumor behavior according to size, histologic characteristics, and level of differentiation, and they found no overall association between the level of comorbidity and BCC risk. In addition, there was no increase in the detection of smaller BCCs that would typically occur if surveillance bias was present.

Selection bias might also be due to the relative triviality of BCC compared with severe chronic conditions. There is therefore a possibility of underdiagnosing BCC in such conditions. Because this problem is most frequent with older populations, potential selection bias among cases being ascertained and cases not being ascertained was evaluated by performing age-group analyses (<61, 61–80, and ≥81 years old). The investigators found that there was a similar risk of BCC in both younger and older age groups, eliminating the possibility of an underestimation of the overall effect of chronic disease on the risk of BCC. An unbiased control group was obtained by selecting controls from the general population.

4. Elevated risk of NMSC was seen in hospitalized patients with connective tissue disease, organ transplants, lymphoma, and severe skin disease.

There was an overall association between SCC and hospitalizations for chronic diseases (IRR 1.47; 95% CI: 1.14–1.91), but this association no longer existed after exclusion of disease categories known to elevate NMSC risk (IRR 1.11; 95% CI: 0.86–1.43). More specifically, there was an elevated risk of SCC in leukemia patients (IRR 7.75; 95% CI: 2.35–26), an elevated risk of SCC in patients with severe skin diseases (IRR 5.28; 95% CI: 1.95-14), and a trend toward an elevated risk of SCC in lymphoma patients (IRR 3.86; 95% CI: 0.99–15). However, the 95% confidence interval for lymphoma included 1.0, suggesting a trend rather than statistical significance.

There was no overall association between BCC and hospitalizations for chronic diseases (IRR 1.03; 95% CI: 0.93–1.14). This remained the case after excluding disease categories known to increase the risk of NMSC (IRR 0.97; 95% CI: 0.90-1.05). In some specific diseases an association was found. More specifically, an elevated risk of BCC was found in transplant patients (IRR 8.00; 95% CI: 2.15–30) and lymphoma patients (IRR 2.50; 95% CI: 1.29–4.84). A trend toward an elevated risk for BCC was found among patients with connective tissue disease (IRR 1.34; 95% CI: 0.99–1.82), because the confidence interval, once again, included 1.0.

5. The major assertion of the investigators is that there is an association between some specific severe chronic diseases that require hospitalization and the incidence of NMSC. They state that a major limitation in their study was the inability to estimate risk precisely when examining specific chronic diseases, particularly because of the relatively small number of SCCs. Because the risk of NMSC was increased among patients with connective tissue disease, organ transplants, lymphoma, and severe skin disease, the investigators suggest a causal effect of general immunosuppression in the development of NMSC. When disease categories affecting the immune system (known to increase the risk of NMSC) are excluded, the association between SCC development and chronic diseases disappears. The investigators state that this is due to a relationship among a chronic disease condition, the immune system, and the development of NMSC. Chronic pulmonary disease was also associated with increased risk for NMSC. This may be due to a similar risk factor (smoking) or the causal effect of a common treatment medication for chronic pulmonary disease (glucocorticoid therapy) that raises the risk for the development of NMSC.

6. The major implication of this study is that patients who have been hospitalized with chronic conditions may need increased surveillance for NMSC. While several mechanisms were proposed for NMSC development, it would be important to discover the mechanisms involved in the increased incidence of NMSC in order to attempt to prevent it. For example, whether persons with lower risk of development of skin cancer—such as those with a higher Fitzpatrick skin type or greater constitutive pigmentation—have an increased risk in the face of these comorbidities is not known, but it would be important information to have when caring for such patients.

7. This study utilized a database to develop a retrospective cohort of patients previously hospitalized for chronic diseases who then developed NMSC. A prospective long-term study of patients with specific chronic conditions and controls followed longitudinally to determine whether they develop NMSC at a greater frequency would be of value. It would be important to control for medications and disease severity. Non-hospitalized patients could also be studied; hospitalization as an independent risk factor could thus be potentially eliminated. The study and control groups would need to be monitored using the same standard and mechanisms to eliminate surveillance bias.

Establishing whether a reciprocal relationship exists between NMSC and specific chronic conditions would require following patients diagnosed with NMSC and controls to determine how often they develop various chronic conditions.

REFERENCES
Agarwal SK, Marshall GD Jr (1998) Glucocorticoid-induced type 1/type 2 cytokine alterations in humans: a model for stress-related immune dysfunction. J Interferon Cytokine Res 18:1059–68

Charlson ME, Pompei P, Ales K, MacKenzie CR (1987) A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 40:373–83

De Hertog SA, Wensveen CA, Bastiaens MT, Kielich CJ, Berkhout MJ, Westendorp RG et al. (2001) Relation between smoking and skin cancer. J Clin Oncol 19:231–8

Frisch M, Melbye M (1995) New primary cancers after squamous cell skin cancer. Am J Epidemiol 141:916–22

Frisch M, Hjalgrim H, Olsen JH, Melbye M (1996) Risk of subsequent cancer after diagnosis of basal cell carcinoma, a population-based, epidemiologic study. Ann Intern Med 125:815–21

Hemminki K, Jiang Y, Steineck G (2003) Skin cancer and non-Hodgkin’s lymphoma as second malignancies: markers of impaired immune function? Eur J Cancer 39:223–9

Hjalgrim H, Frisch M, Storm HH, Glimelius B, Pedersen JB, Melbye M (2000) Non-melanoma skin cancer may be a marker of poor prognosis in patients with non-Hodgkin’s lymphoma. Int J Cancer 85:639–42

Hönigsmann H, Wolff K, Gschnait F, Brenner W, Jaschke E (1980) Keratoses and nonmelanoma skin tumors in long-term photochemotherapy (PUVA). J Am Acad Dermatol 3:406–14

Hu S, Federman DG, Ma F, Kirsner RS (2005) Skin cancer and non-Hodgkin’s lymphoma: examining the link. Dermatol Surg 31:76–82

Jensen AO, Olesen AB, Dethlefsen C, Sørensen HT (2006) Ten year mortality in a cohort of nonmelanoma skin cancer patients in Denmark. _J Invest Dermatol_126:2539–41

Jensen AO, Olesen AB, Dethlefsen C, Sørensen HT, Karagas MR (2008) Chronic diseases requiring hospitalization and risk of non-melanoma skin cancers—a population based study from Denmark. J Invest Dermatol 128:926–31

Karagas MR, Cushing GL Jr, Greenberg ER, Mott LA, Spencer SK, Nierenberg DW (2001) Non-melanoma skin cancers and glucocorticoid therapy. Br J Cancer 85:683–6

Lutzner MA (1984) Skin cancer in immunosuppressed organ transplant recipients. J Am Acad Dermatol 11:891–3

Martin H, Strong E, Spiro RH (1970) Radiation-induced skin cancer of the head and neck. Cancer 25:61–71

Saladi RN, Persaud AN (2005) The causes of skin cancer: a comprehensive review. Drugs Today (Barc) 41:37–53

¹Department of Dermatology and Cutaneous Surgery, University of Miami Miller School of Medicine, Miami, Florida, USA