Schematic overview of study population. BCC indicates basal cell carcinoma.
The absolute basal cell carcinoma count in the 10 820 cohort members (mean [SD] tumor count, 1.63 [1.348]).
Time to development of subsequent basal cell carcinoma lesions. Zero indicates multiple lesions present at index date.
Kiiski V, de Vries E, Flohil SC, Bijl MJ, Hofman A, Stricker BHC, Nijsten T. Risk Factors for Single and Multiple Basal Cell Carcinomas. Arch Dermatol. 2010;146(8):848-855. doi:10.1001/archdermatol.2010.155
To investigate the incidence of single and multiple basal cell carcinoma (BCC) lesions and associated risk factors.
A prospective, population-based cohort study (from January 1, 1990, through December 31, 2007).
Two cohorts of 10 994 Dutch people, 55 years or older, were studied in 1990 (first cohort) and 1999 (second cohort).
Patients with BCC lesions were identified from the Dutch national pathology laboratories network, hospitals, and general practices.
Main Outcome Measures
The associations between determinants and single and multiple BCC lesions were studied by estimating odds ratios (ORs) and hazards ratios, using multivariate logistic regression and Andersen-Gill models, respectively.
Of the eligible 10 820 cohort members, 524 (4.8%) had BCC, of whom 361 had single and 163 (31.1%) had multiple lesions. Age and red hair were significant risk factors for a first BCC lesion in a multivariate model. In the Andersen-Gill model, people who developed a first BCC lesion after 75.0 years of age were significantly less likely to develop multiple lesions (≥75.0 years adjusted OR, 0.58; 95% confidence interval [CI], 0.47-0.71). Red hair (adjusted OR, 1.43; 95% CI, 1.05-1.94), high educational level (1.42; 1.12-1.81), and a first BCC lesion located on the upper extremities (1.49; 1.02-2.15) were associated with a significantly increased risk of developing multiple lesions.
Patients who are relatively young at their first BCC diagnosis, those with red hair, those with higher socioeconomic status, and/or those with a BCC lesion on their upper extremities have a higher risk of developing multiple lesions and require closer follow-up over time.
Basal cell carcinoma (BCC) is the most common type of cancer in people with European ancestry, and its incidence continues to increase steeply.1- 5 The incidence of BCC varies geographically. For instance, in the Netherlands, the incidence rate is approximately 100 per 100 000 person-years, which is approximately one-tenth the risk in areas such as Australia.6 More than a million people in the United States develop BCC annually.7 Although BCC therapy is relatively straightforward and BCC mortality rates are extremely low, the high incidence of BCC and the high risk of developing multiple lesions put a major burden on limited health care resources, placing BCC in fifth place on the list of the most expensive cancers to treat in the United States.8
Individual risk factors for BCC include age, male sex, race, phenotypic characteristics, and genetic predisposition. These factors may interact with environmental exposures, such as UV light or iatrogenic exposures.9- 14 In contrast to risk factors associated with incident BCC, the risk factor profile of those who develop multiple BCC lesions among patients with a prior lesion is not well documented. Most observational BCC studies were performed with data from a few cancer registries that record incident BCC without detailed data on risk factors (except basic demographics). Specialized centers include specific subgroups of patients with detailed risk factor information, but their results are often not generalizable because of selection processes with potential bias or confounding.12,15- 17 The strongest risk factor for developing multiple BCC lesions appears to be a history of a prior lesion, representing the accumulation of gene-environmental interactions. An estimated 40% to 50% of patients with a BCC lesion develop subsequent tumors.12,13 Previous studies12,16- 18 have suggested that patients with BCC who had a truncal and/or superficial lesions, who have a sun-sensitive skin type, and who are unable to tan are at increased risk of developing subsequent tumors.
Identification of patients with BCC at high risk of developing subsequent lesions may assist physicians in the adequate selection of individuals from the large number of patients with BCC who should be followed up more closely over time. Currently, national BCC guidelines are ambiguous concerning follow-up but often advise patients to be followed up annually for several years (often a 5-year period). This recommendation leads to an overwhelming workload for dermatologists, of whom there is a shortage in several countries, such as the Netherlands, the United States, and the United Kingdom.19- 21 Therefore, the study objective was to investigate incidence and risk factors associated with the development of single and multiple BCC lesions in a population-based study of almost 11 000 people aged 55 years or older from the general Dutch population.
The Rotterdam Study is a well-established, prospective, population-based cohort study based in the Ommoord district in Rotterdam, the Netherlands; it was designed to study diseases in elderly people.22 In January 1990, the first cohort of 7983 people 55 years or older (78.1% of invitees) was established. In 1999, an additional 3011 participants (67.3% of invitees) who had turned 55 years old, or were already of the target age range and had moved into the district, were added to the cohort. The Medical Ethics Committee of the Erasmus Medical Center approved the Rotterdam Study; written informed consent was obtained from each participant.
All pathology report–confirmed cases of BCC in the Rotterdam Study were extracted from the Dutch national pathology laboratories network (Pathologisch Anatomisch Landelijk Geautomatiseerd Archief) between the start of the cohort study on January 1, 1990, and December 31, 2007. Patients with a BCC diagnosis before study entry were excluded from the analyses (Figure 1). The BCC reports with pathological sample dates 6 months or less apart, with the same physical location and/or pathology report summary including the words biopsy or punch biopsy at first date and excision on the second or following dates, were considered to be the same BCC occurrence and therefore counted as 1 lesion. If no further description was available, BCC reports in cases appearing within 3 months in the same physical location were combined, assuming that the data specified a biopsy followed by an excision. Subsequent lesions occurring in the same location with recurrence or reexcision mentioned in the pathology report were considered to be recurrent tumors and were not counted as subsequent BCC lesions. Lesions occurring within 6 months of the first date of BCC diagnosis (index date) were counted as additional tumors at the date of the first diagnosis (index lesions) because they probably were also present at the index date. These lesions were counted separately, with an interval of zero between occurrences, compared with the index lesion. Tumor-specific information was extracted from the Systemized Nomenclature of Medicine abstract available in the Dutch national pathology laboratories network.
The anatomical site of the BCC lesion was classified by dividing the surface area of the body into 20 parts. Histologic subtypes were categorized as nodular (including nodulocystic and basosquamous types), infiltrative, superficial, micronodular, and adenoid. In mixed-type lesions a superiority rule was used according to aggressiveness, namely, infiltrative greater than nodular greater than micronodular greater than superficial. When inconsistent, the pathology report of the excision overruled the biopsy report. The BCC count was continuous; multiple BCC lesions were defined as 2 or more lesions. Patients were counted only once per histologic subtypes or anatomical sites in analyses that stratified for these 2 characteristics. Cohort participants without a pathology report–based diagnosis of BCC were used as a reference group.
Baseline data of determinants and potential risk factors for BCC included sex, age, hair color when young (fair/blond, red, or brown/black, as determined by self-report), eye color (blue, brown, or intermediate [defined as neither blue nor brown]). Also, educational level (completion of low [primary education (6 years)], medium [lower-level general education (10 years), intermediate-level general education (11 years), or lower-level vocational education (10 years)], or high [higher-level general education (12 years), intermediate-level vocational education (14 years), or higher-level vocational education (15 years) and university (16-18 years)]), alcohol consumption (<10, 10-20, or >20 g/d), smoking history (never, former, or current smoker), and body mass index (calculated as weight in kilograms divided by height in meters squared; <25.00, 25.00-29.99, or ≥30.00) were factors determined at cohort entry. Four questions assessing UV exposure were available, including tendency to develop sunburn, history of more than 25 years of outdoor work, having lived more than 1 year in a sunny country (countries not specified; determined by self-report), and sun-protective behavior (ie, wearing sunglasses and/or a wide-brimmed hat when in the sun; determined by self-report). The first 3 UV items had binary responses, and the fourth had the possible responses of no, often, and always.
Differences in the distribution of demographic and BCC characteristics were compared using the t-test or Mann-Whitney test for continuous variables, as appropriate, and the Pearson χ2 test for categorical variables. Because of violation of the proportional hazards assumption, logistic regression models were used to calculate crude and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) for the development of a single or multiple BCC lesions. In addition to sex, if a variable showed an association with BCC occurrence with P < .20 in the univariate analyses, it was included in the multivariate logistic regression model.23 No significant interactions were observed among variables that were eligible for the multivariate logistic regression model.
The Andersen-Gill multifailure survival model, which assesses the recurrence of multiple events, was used to estimate hazards ratios (HRs) and 95% CIs for multiple BCC lesions. The advantage of this method of analysis is that it includes all BCC lesions patients developed after the initial lesion, not just the second, which makes it more appropriate for depicting a phenomenon of a repeated character, such as BCC occurrence.15 To increase the stability of the Andersen-Gill analysis, patient data were censored at a maximum of 5 lesions because of the small number of patients with more than 5 (n = 12). Variables showing an association with P < .20 were included in the adjusted model. The HRs for lesion subtype and lesion site were analyzed by making dummy variables by subcategory, resulting in analyses comparing 1 subtype or site with all other subtypes or sites as a reference group. SPSS statistical software, version 15.0 for Windows (SPSS Inc, Chicago, Illinois), was used for all data analyses except the Andersen-Gill multifailure survival model, for which SAS statistical software, version 9.13 for Windows (SAS institute Inc, Cary, North Carolina), was used.
Of the 10 994 persons who participated in the Rotterdam Study, 174 (1.6%) were excluded because they had a BCC lesion before cohort entry (Figure 1). The remaining 10 820 patients were included; they contributed 102 171 person-years of follow-up, with a mean (SD) follow-up duration of 9.5 (4.8) years. The mean age at study entry was 69.1 years (Table 1). Of our study population, 59.9% was female and 97.9% was white. Characteristics of the study population are presented in Table 1.
A total of 1556 pathology reports containing information on BCC lesions of the cohort members of the Rotterdam Study from January 1, 1982, through December 31, 2007, were extracted from the Dutch national pathology laboratories network. Of the 10 820 eligible cohort members, 524 (4.8%) had at least 1 recorded pathology report–confirmed lesion, of whom 361 (68.9%) had one, 95 (18.1%) had two, and 68 (12.9%) had three or more tumors (Figure 2). The mean and median numbers of pathology report–confirmed lesions were 1.63 and 1, respectively (range, 1-11). In total, 854 unique lesions (recurrences excluded) were identified, of which approximately two-thirds were located on the head and neck (Table 2). More than half the lesions were nodular and about a quarter were of infiltrating histologic subtype. Of all patients with BCC, 335 (64.9%) had at least 1 cancerous lesion of the nodular subtype, 165 (31.5%) had infiltrating lesions, and 77 (14.7%) had superficial lesions.
In the group of patients with multiple BCC lesions, 129 (79.4%) had 1 or more nodular lesions, 71 (43.6%) had infiltrating lesions, and 50 (30.9%) superficial lesions. Most lesions were located in the head and neck area (298; 60.4%), followed by the trunk and upper and lower extremities (88 [17.8%], 26 [5.3%], and 57 [11.6%] of all lesions, respectively). Among patients with multiple lesions, almost half (37; 48.1%) of those who had a superficial lesion also had at least 1 nodular lesion and 15 (19.5%) also had an infiltrating lesion.
During the first year of follow-up, 0.2% of cohort members developed a BCC lesion; cumulative percentages of developing BCC during 3, 5, and 10 years of follow-up were 0.9%, 1.8%, and 3.4%, respectively. The incidence rate of a first lesion was 513 per 100 000 person-years. The incidence among women (497 per 100 000 person-years) was lower than that among men (566 per 100 000 person-years); the risk difference of 87 per 100 000 person-years was not statistically significant (P = .06).
The incidence of multiple BCC lesions was 161.5 per 100 000 person-years, and it was significantly lower for women than for men (139 vs 197 per 100 000 person-years, respectively; risk difference of 58 per 100 000 person-years; P = .02). Among those with a first lesion, 3-year, 5-year, and 10-year incidence rates per 100 000 person-years for developing multiple lesions were 893, 1459, and 1974, respectively. Of the patients with multiple lesions, 140 (85.9%) developed their second lesion during a follow-up period of 5 years after the first lesion, indicating that the development of a subsequent lesion is more likely in the short-term period after the first lesion than in the late-term period (>5 years). Furthermore, 70 (42.9%) of the patients had multiple lesions on the date of diagnosis (Figure 3).
In univariate analyses, an age of 65.0 to 75.0 years at study entry, alcohol consumption, fair/blond or red hair, and high tendency to develop sunburn were significantly associated with increased risk of developing a first BCC lesion (Table 1). Red hair had the strongest association (OR, 1.98; 95% CI, 1.31-3.01). Eye color, educational level, outdoor work history, history of living in a sunny country, sun-protective behavior, and body mass index were not significantly associated with higher risk of BCC. In univariate analysis, current smoking was associated with decreased risk of being diagnosed as having BCC (OR, 0.77; 95% CI, 0.59-1.00).
After including sex, age, hair and eye color, tendency to develop sunburn, history of outdoor work, living in a sunny country, smoking history, alcohol consumption, body mass index, and educational level (P < .20 in the univariate analyses) in a multivariate logistic regression model, older age at study entry (65.0-74.9 years; adjusted OR, 1.39; 95% CI, 1.11-1.75; compared with age <65.0 years) and red hair (adjusted OR, 1.98; 95% CI, 1.24-3.14) remained significantly associated with increased risk of developing a first BCC lesion. In the multivariate model, a high tendency to develop sunburn was no longer a significant risk factor, but medium educational level was positively associated with increased risk.
In the univariate Andersen-Gill multifailure analyses, several demographic factors (eg, male sex, red hair, and high tendency to develop sunburn) and tumor characteristics, as well as higher educational levels, were significantly associated with increased risk of developing multiple BCC lesions. In contrast, older age at the time of the first lesion and alcohol consumption were associated with a decreased risk of subsequent lesions (Table 3). In the univariate analyses, superficial histologic type at first occurrence of BCC and lesions located other than in the head and neck region seemed to increase the likelihood of developing subsequent lesions.
In the multivariate model that adjusted for age at index lesion, sex, hair color, eye color, all UV-related factors, smoking history, alcohol consumption, lesion histologic subtype and location, and age younger than 65.0 years at the time of first BCC lesion were associated with a significantly increased risk of developing multiple lesions (eg, ≥75.0 vs 65.0-74.99 years; adjusted HR, 0.58; 95% CI, 0.47-0.71; and HR, 0.65; 95% CI, 0.53-0.81, respectively; Table 3). Older age (age groups of 65.0-74.99 and ≥75.0 years) at index lesion is associated with a decreased risk of developing multiple lesions. For the oldest age group, this finding could be partly explained by the mean follow-up time, which is relatively shorter than that of the middle and youngest age groups (mean follow-up times, 6.66, 10.36, and 10.70 years, respectively). However, this theory does not clarify the decreased risk of developing multiple lesions in the 65.0-year to 74.99-year age group. Blue eyes and fair/blond hair were not significantly associated with developing multiple lesions, but people with red hair were 40.3% more likely to develop multiple lesions than those with brown hair (adjusted HR, 1.43; 95% CI, 1.05-1.93). Compared with patients with low educational level, those with a medium or high educational level were 20.2% and 40.2%, respectively, more likely to develop multiple lesions. After adjusting for confounders, the index lesion location on the upper extremities was the only BCC characteristic that remained significantly associated with developing multiple lesions (adjusted HR, 1.49; 95% CI, 1.02-2.15).
In this study, 524 (4.8%) of the cohort members had a BCC lesion and approximately one-third of them developed multiple lesions during a mean of almost 10 years of follow-up; these findings are consistent with data from other studies.12,13,16 Most risk factors associated with developing a first lesion, such as age and hair color, are in accordance with findings from other studies, although men in our study were not found to be at increased risk, which has been observed in some but not all previous studies.12,17 As expected, BCC risk increased with age, but people who developed their first lesion before 65.0 years of age were significantly more likely to develop multiple lesions. In contrast to developing a first lesion, high educational level was significantly positively associated with developing multiple lesions. This finding may be explained by the probability that people with higher levels of education (which correlates strongly with socioeconomic status) have different lifestyles (eg, more frequent exposure to UV rays for intermittent periods). It may also be partly explained by findings that these people are more likely to develop superficial lesions in regions other than the face and neck (these lesions are associated with the risk of developing multiple lesions), or because of the well-known time bias (eg, people with higher levels of education and socioeconomic status are more likely to live longer; thus, they may have more time to develop lesions).24
However, the mean age at study entry of the participants with a higher educational level was younger (64.77 vs 71.32 years) and the mean follow-up period shorter than in the group of participants with only completion of primary education (7.75 vs 10.47 years), thereby suggesting that this bias does not affect our findings. Interestingly, after adjusting for confounding factors, known BCC risk factors, such as blue eyes and fair/blond hair, were not associated with developing multiple lesions. Red hair seemed to be the most important risk factor for developing subsequent lesions. No associations were observed between BCC and variables assessing, in part, cumulative UV exposure during one's lifetime (eg, sun-protective behavior, outdoor work, and history of living in a sunny country).
The observed discrepancy among risk factor profiles of developing single or multiple BCC lesions may suggest that once cumulative environmental-genetic interaction has surpassed a certain threshold and resulted in a lesion, the phenotypic characteristics of patients seem less important. The clinical relevance of this finding is that physicians' risk assessment efforts should differentiate between patients at risk for a first lesion and those who have a history of BCC. Of the people with prior BCC, those who are relatively young at diagnosis of a first lesion, those who have red hair and higher socioeconomic status, and/or those who had a lesion on their upper extremities may require a more stringent follow-up regimen than other patients with BCC. However, of the classic phenotypic, UV exposure, and lifestyle risk factors examined in this study, the strength of the risk estimates was modest (adjusted HR, <1.5), suggesting that other (genetic) factors may play an important role in the predisposition for developing multiple lesions. In this sample of the general population, more than 30% of the patients with BCC developed subsequent skin cancer, emphasizing the need for (annual) follow-up for several years. This recommendation is often stated in the Dutch national BCC guidelines (which are also followed by other countries, including Australia and the United States) and has major implications in the allocation of dermatologic care because of the enormous volume of skin cancer patients in predominantly white populations (most notably in Australia and the United States but also in Europe).
Interestingly, more than half the additional BCC lesions among patients with at least 1 superficial lesion were of the nodular or infiltrating histologic type, suggesting a common pathogenesis of these clinically and histologically different types of BCC. In clinical practices, this implies that patients with superficial lesions, which are not very aggressive, are at high risk of developing more aggressive high-risk lesions.25 We confirmed that a lesion in a location other than the head or neck is a risk factor for developing multiple lesions, especially ones located on the upper extremities.16,26 Most lesions were located on the head and neck area, followed by the trunk and upper and lower extremities; these findings are in accordance with those of previous studies27 concerning anatomical distribution.
To our knowledge, ours is the largest population-based study (including 100 000 person-years of follow-up) to assess and compare risk factor profiles of first and multiple BCC lesions based on detailed information with regard to BCC and patient characteristics. In total, 4 UV-related questions were asked (Tables 1 and 3); however, no information was available concerning UV exposure in childhood and adolescence, which is a limitation because this information seems to be important in the pathogenesis of BCC. The study population consisted only of people 55 years or older, leading to a small overrepresentation (6481; 59.9%) of women in the study population. Because only histologically confirmed lesions were included in this study, those not biopsied or treated without histopathologic confirmation were missed. However, in the study period, the recommended treatment according to the Dutch BCC guidelines was surgical excision; the community hospital responsible for most of the health care provided in the study region usually required pathological confirmation of diagnosis and used relatively few noninvasive skin cancer therapies (cryotherapy, radiotherapy, topical therapies, and photodynamic therapy) during the study period (oral communication, Marinus van Praag, MD, PhD, September 7, 2009). This information suggests that the number of missed BCCs was limited.19 To minimize multiple counting of a single lesion, a robust and conservative attempt was made to differentiate among biopsied, excised, incident, recurrent, and subsequent lesions.
In conclusion, 30.9% of Dutch patients with BCC developed multiple lesions. The risk profiles associated with incident and multiple lesions differed; this information requires physicians to alter their risk assessment after a first lesion has been diagnosed. More research is needed to identify people who are at risk of developing multiple lesions because the follow-up of this large group of people is putting a strain on limited specialized care.
Corresponding Author: Bruno H. C. Stricker, MD, PhD, Department of Epidemiology, Erasmus Medical Center, PO Box 2040, 3000 CA Rotterdam, the Netherlands (firstname.lastname@example.org).
Accepted for Publication: January 4, 2010.
Author Contributions: Drs Kiiski, de Vries, and Nijsten had full access to all the data and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Stricker and Nijsten. Analysis and interpretation of data: Kiiski, Flohil, Bijl, de Vries, Stricker, and Nijsten. Drafting of the manuscript: Kiiski and Nijsten. Critical revision of the manuscript for important intellectual content: de Vries, Flohil, Bijl, Hofman, Stricker, and Nijsten. Statistical analysis: Kiiski, de Vries, Flohil, and Bijl. Administrative, technical, and material support: Hofman and Stricker.
Financial Disclosure: None reported.