Although sun exposure is known to be associated with basal cell carcinoma (BCC), it is not known what determines multiple occurrences of BCCs among sporadically affected individuals or why BCCs develop on uncommonly sun-exposed body sites like the trunk. In a prospective community-based skin cancer study in Queensland, Australia, we studied all participants who experienced a histologically confirmed BCC from 1992 to 2007. Sun exposure history was monitored, and dermatologists documented phenotype at baseline and signs of photodamage over the study period. Anatomic sites of all incident BCCs were recorded. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated using logistic regression. Of 401 participants who developed a new BCC during the 16 years of follow-up, 232 (58%) developed more than 1. Male sex (OR = 2.5, 95% CI 1.5–5.3) and age 60 or over (OR = 4.2, 95% CI 1.5–11.8) but not skin type were associated with highest BCC counts among those affected. Participants with high numbers of solar keratoses were most likely to experience the highest BCC counts overall (OR = 4.3, 95% CI 1.4–13.5). Moreover, occurrences of BCC on the trunk (OR = 3.3, 95% CI 1.4–7.6) and on the limbs (OR = 3.7, 95% CI 2.0–7.0) were strongly associated with high numbers of solar keratoses on these sites, respectively. Among those newly affected by BCC, chronic cutaneous sun damage predicts those who will be affected by more than 1 BCC, while chronic sun damage on the trunk and limbs predicts BCC occurrence on the trunk and limbs, respectively.
Basal cell carcinoma (BCC), the commonest cancer, occurs more than once in a high proportion of individuals sporadically affected. In North American study populations with relatively low incidence of BCC,1, 2 2 prospective cohort studies reported that around 34% of affected people developed multiple BCCs within 5 years of follow-up.3, 4 In a high-incidence Australian population, nearly 50% of affected people developed multiple BCCs over 10 years.5 Hence, it is not only the number of people affected but also the multiple tumors they develop that underlies the substantial health expenditure attributed to BCC treatment in North America,6 Europe7 and Australia.8
A limited number of previous studies in selected clinical populations have suggested that multiple BCC occurrence is associated with increasing age,9 male sex10 and fair skin,11, 12 but little is known about the importance of other pigmentary and environmental factors including sun exposure history and photodamage.
Another unanswered question about BCC is why a sizable proportion occurs on sites that are often sun protected like the trunk and limbs, given BCC's strong association with sun exposure.13 Some have suggested different mechanisms of BCC pathogenesis on different sites because head/neck BCCs appeared more strongly associated with a sun-sensitive phenotype than trunk BCCs.12, 14, 15 Also, people initially affected by trunk BCC were reported to be affected more often subsequently on the trunk than people with an initial head/neck BCC.16 Overall, however, there is limited evidence about where on the body successive BCCs arise among people affected more than once,9, 16, 17 and factors underlying BCC's site distribution in general remain uncertain.
To answer these questions, we monitored the occurrence of all BCCs in a subtropical Australian community from 1992 to 2007. We examined the anatomic sites of BCC occurrence and assessed in detail the associations between pigmentary traits and various measures of sun exposure with BCC multiplicity. We hypothesized that people who develop multiple BCCs would have a more sun-sensitive phenotype than people with a single BCC during the same period.
Material and Methods
The Nambour skin cancer study
Study participants were originally randomly selected in 1986 from the electoral register of all adult residents of the subtropical township of Nambour in Queensland, Australia (latitude 26°S) for a baseline study of skin cancer.18 A total of 2,095 people (70%) took part and were shown to be a representative sample of the community with regard to risk factors for skin cancer.18 Of these, 1,621 (77%) participated in a field trial between 1992 and 1996 to assess sunscreen application and betacarotene supplementation in the prevention of skin cancer. Detailed descriptions of the community sample, field trial and its long-term outcomes have been published previously.19, 20 On conclusion of the trial in 1996, participants were invited to take part in a further follow-up study of skin cancer. Continuing participants consented to have subsequently diagnosed skin cancers notified to the investigators by regional pathology laboratories in Queensland, allowing 100% ascertainment of all histologically confirmed BCCs. No participants were immunocompromised, and none had nevoid BCC syndrome. Follow-up of skin cancers continued to the end of 2007. Ethical approval was obtained from the ethics committee of the Queensland Institute of Medical Research.
Pigmentary traits, sun exposure and photodamage.
Participants' head, neck, arms and hands were examined at baseline in 1986 by dermatologists who documented standard signs of cumulative photodamage (telangiectasia of the face, elastosis of the neck, solar lentigines and solar keratoses) and number of melanocytic nevi on arms. All participants completed a baseline standard questionnaire about pigmentary traits (hair and eye color) and sun exposure habits (occupational and recreational, and sunscreen use when in the sun). At the start of the trial in 1992, participants again completed a standard questionnaire including their skin response to acute sun exposure. In 1992, 1994 and 1996, information about sun exposure was updated by questionnaire, and dermatologists carried out full skin examinations. From 1997 to 2007, participants reported their sunscreen use each year and based on their responses were classified as regular or irregular sunscreen users.21
Basal cell carcinomas.
All BCCs were verified histologically. Because the same BCC may be histologically confirmed twice, at initial biopsy/excision and again if reexcised, all records of apparently multiple BCCs verified within a 6-month period in the same person and on the same anatomic site were cross checked to identify duplicate reports. Recurrent BCCs diagnosed at the sites of earlier primary lesions were excluded.
Cases and tumors.
Only people who experienced a new histologically diagnosed BCC during the study period, 1992–2007, were included in the analyses. For investigation of the number of BCCs that participants developed, analyses were person based, and affected participants were classified according to the number of BCCs they developed in the 16-year study period: 1, 2–3 and 4 or more (synchronously and/or over time). Anatomic site analyses were lesion based such that for analyses of associations with BCCs occurring on the head/neck, all BCCs that occurred on the trunk or limbs constituted the reference group. Thus, although the sites of occurrence of individual BCCs were mutually exclusive, the sites of BCC occurrence within multiply affected individuals were not, in the sense that they could contribute BCCs to several sites.
For person-based analyses, only sun exposure and photodamage data before the date of diagnosis of an individual's first BCC were used; for tumor-based analyses, all exposures predated diagnosis of each BCC when more than 1 BCC occurred. For example, for a BCC diagnosed in 1995, the extent of photodamage was taken as the greatest recorded in either 1986, 1992 or 1994 (but not 1996). The responses used for recreational and occupational sun exposure analyses always reflected the level of these exposures reported for the majority of time before BCC diagnosis.
Multivariate polytomous and repeated-measures binomial logistic regression were used to assess the associations between factors of interest and BCC multiplicity and site of occurrence among those affected, respectively. The repeated-measures analyses accounted for the intraperson correlation of BCCs on different sites from the same person. Each exposure variable was treated categorically to estimate the odds ratio (OR) with 95% confidence intervals. All analyses were adjusted for age, sex and randomized sunscreen treatment during the trial. Models also included pigmentary traits and sun exposure variables where appropriate. Statistical analyses were conducted using SAS version 9.1 (SAS Institute, Cary, NC) and SUDAAN 10 (RTI Institute, Research Triangle Park, NC).
During the 16-year observation period, 401 participants (30%) of the overall Nambour study cohort had at least 1 surgically excised, histologically confirmed BCC. Of these 401, 66 had a history of BCC before 1992. During the study period, approximately equal proportions of women (207; 52%) and men (194; 48%) were diagnosed with an incident BCC, with an average age at diagnosis of the first BCC in the study period of 60 years for women and 61 years for men. The average age at diagnosis of a first BCC was the same for those who developed 1 or more than 1 BCC (mean age 60 years). As anticipated, a high proportion of cases had a sun-sensitive phenotype: 62% had fair skin color, 49% had blue/gray eyes and 60% had a light hair color (Table 1).
Table 1. Odds ratios for multiple basal cell carcinomas (BCC) compared to people with a single BCC in 1992–2007, according to age, sex, pigmentary traits, sun exposure and cutaneous signs of sun damage in Queensland, Australia
Among those with a new BCC in the 16-year period, 169 people developed 1 BCC only (42%; 100 women and 69 men) and 232 people developed more than 1 BCC (107 women and 125 men), representing a cumulative incidence of multiple BCC of 58%. Among those people who experienced more than 1 BCC, 61 (26%) had 3 BCC tumors, 31 (13%) had 4 BCC tumors and 13 (6%) had 5 or more (up to 37) BCC tumors. A total of 1,202 new BCCs were excised from the 401 affected individuals between 1992 and 2007.
Associations with BCC multiplicity among those affected by BCC
People with multiple BCCs in the study period, synchronously and/or over time, were more likely to be older and male than those with a single BCC. Specifically, those aged 60 years or older at the start of observation were more likely to develop ≥4 BCCs (OR 4.2, 95% CI 1.5–11.8) compared to people younger than 40 years. Males were more likely than females to be affected by 2–3 BCCs (OR 1.8, 95% CI 1.1–3.2) and ≥4 BCCs (OR 2.8, 95% CI 1.5–5.3) among all affected by BCC in the study period (Table 1).
Compared to cases who tanned, cases who had skin that always burnt upon exposure to the summer sun were not more likely to develop multiple BCCs (Table 1). Cases with light eye color were more likely to develop multiple BCCs compared to cases with dark eyes, though this was significant only for people with hazel/green eyes who developed 2–3 BCCs (OR 2.5, 95% CI 1.0–6.5). Although presence of melanocytic nevi on the arms was not associated with multiple BCCs, cases with nevi on the back were significantly less likely to develop multiple BCCs (2–3 BCCs: OR 0.39, 95% CI 0.18–0.84, ≥4 BCCs OR 0.29, 95% CI 0.12–0.67, Table 1). No association was found between recreational or occupational sun exposure and BCC multiplicity (Table 1).
Regarding measures of acute photodamage, number of lifetime sunburns reported before BCC diagnosis was not associated with multiplicity of BCC among cases (Table 1) and nor were facial telangiectasia and solar elastosis of the neck, as measures of chronic photodamage (Table 1). Among all those affected by BCC, strong associations with increasing BCC multiplicity were evident for increasing total body SK counts such that for those with 5 or more SKs the odds for developing ≥4 BCCs over the observation period were 4 times greater (OR 4.3, 95% CI 1.4–13.5) compared to those with no SKs (Table 1).
Pattern of anatomic sites affected by BCC
Among the 169 study participants affected only once by BCC in the study period, nearly half were affected on the head/neck (47%) followed by the trunk or limbs (29 and 24%, respectively; Fig. 1). Among the 232 participants with more than 1 BCC, 140 people (60%) were affected on more than 1 anatomic site. Of the 139 who developed 2–3 BCCs, the most common site combination within an individual was both the head/neck and trunk affected. However, among the 93 who developed ≥4 BCCs, the tumors usually occurred across all the 3 anatomic sites specified (Fig. 1). Among cases multiply affected, 192 (83%) people experienced at least 1 BCC on the head/neck region, 126 (54%) at least 1 BCC on the trunk and 75 (32%) at least 1 BCC on the limbs. Overall, the pattern of tumor occurrence across anatomic sites was similar, whether cases had a single BCC or multiple BCCs, with the head/neck by far the most commonly affected site (Fig. 1).
Associations with anatomic site of BCC
Among all newly affected by BCC in the period, neither age nor sex was significantly associated with the anatomic site of BCC development (Supporting Information Table 1).
Cases with facial telangiectasia were more likely to develop BCCs on the head/neck (OR 2.1, 95% CI 1.0–4.1) compared to cases with no facial telangiectasia (Table 2). The presence of many solar lentigines on the face was also associated with BCC on the head/neck (OR 1.8, 95% CI 1.0–3.2), but solar elastosis of the neck was not (Table 2). Cases with 5 or more SKs on the face were more likely to have BCC on the head/neck (OR 2.0, 95% CI 1.1–3.7) compared to cases without SKs on the face. Similarly, cases with increasing numbers of SKs on the trunk were increasingly likely to develop BCCs on the trunk so that the odds for developing a BCC on the trunk for those who had 5 or more SKs at this body site were 3 times greater (OR 3.3, 95% CI 1.4–7.6) than those without SKs on the trunk. The presence of any solar lentigines on the back was significantly associated with BCCs on the trunk, and the association was strongest among those cases who had many solar lentigines (OR 2.5, 95% CI 1.3–4.9, Table 2), but neither freckles on the trunk nor melanocytic nevi on the back were associated with BCCs on the trunk. Similarly, although melanocytic nevi on the arms were not associated with limb BCCs, 5 or more SKs on the limbs were strongly associated (OR 3.7, 95% CI 2.0–7.0, Table 2). Supporting Information Table 1 shows associations between each measure of photodamage and BCCs at each anatomic site.
Table 2. Odds ratios1 for anatomic site of basal cell carcinoma (BCC) according to cutaneous signs of sun damage of skin at the same site among people affected 1992–2007, in Queensland, Australia
We repeated analyses after exclusion of the 66 participants who had a past history of BCC, most of whom (79%) went on to develop multiple BCCs in the study period. When analyses were based only on those participants (n = 335) who experienced their first BCC during the observation period, there was no material difference in results.
In this prospective study of 401 people affected by incident BCC in a 16-year period, we found that those who developed high numbers of incident BCCs were significantly more likely than those who developed a single BCC to have been diagnosed previously with SKs. However, there were no significant differences in pigmentary traits or self-reported sun exposure histories among BCC cases who developed multiple compared to a single BCC, in contrast to our study hypothesis. BCCs were more likely to occur on sites that had clinical evidence of photodamage than on sites without. We also showed that people with multiple BCCs are usually affected at more than 1 site, almost always including the head/neck.
Previous studies have described the predilection for sun-sensitive pigmentary traits seen among people affected by BCC compared to those without BCC.22–24 Our study extends this work to describe the differences between cases who developed intermediate or high numbers of BCCs compared to those who developed a single BCC in over a decade and a half of follow-up. Complexion type was not associated with the numbers of BCCs people developed, in agreement with a hospital-based British study of 266 BCC cases12 who found no significant association between the skin's ability to tan and rate of development of subsequent BCCs.
We next assessed if sun exposure history was able to distinguish between people with single versus multiple BCCs. However, after prolonged follow-up, we neither found association with recreational or occupational sun exposure nor found any evidence that the number of sunburns, a measure of acute sun damage, was associated with multiple BCCs. Our results again agree with those of the British study, which found no association between individuals' hours of sun exposure per year before their first BCC or number of childhood sunburns and the number of subsequent BCCs.12 Although ultraviolet radiation (UVR) is the main environmental factor in the development of BCCs,13 current available evidence cannot reliably identify the specific aspects of UVR exposure concerned with the occurrence of BCCs (whether initial or subsequent). This can be attributed partly to the recall error that is inherent in all self-reported estimates of past sun exposure and partly to the self-selection of susceptible people to an indoor lifestyle.22
In contrast, an association between objective clinical markers of photodamage and general BCC occurrence was noted, and this has been described previously by other investigators.24, 25 We have now precisely quantified the relationship, in particular the strong associations between total body SK counts and intermediate and high numbers of BCCs, overall and site specifically. The objective presence of SKs flags a history of chronic sun exposure26 and removes the error when recall is used to assess the past sun exposure of an individual. Moreover, their site-specific prevalence can be regarded as evidence for and personal dosimeters of chronic sun damage to the site affected by SKs, making them useful for quantifying site-specific associations between UVR and BCC. Thus, our data indicate that those who develop multiple BCCs have received a greater dose of UVR than people with a single BCC, in general as well as on specific sites, even including the trunk and limbs.
The theory that different causal mechanisms are responsible for the pathogenesis of BCC on different anatomic site was largely based on cross-sectional findings showing that sun sensitivity and cutaneous sun damage were risk factors for a first BCC on the head/neck, whereas a less sun-sensitive phenotype and a history of sunburns were associated with a first BCC on the trunk.12, 14, 15 It is important to note that these studies did not take into account any subsequent BCCs (and their site) that the participants developed. Meanwhile, another study described in broad terms how people with multiple BCCs tended to develop their BCCs on a single site16 presented no evidence about the proportion of people who developed BCC on a single or more than 1 site. On the other hand, a hospital-based Dutch study with a median follow-up time of 3.1 years revealed that 33% of people affected by multiple BCCs were affected across different sites.9 In a US study, 45% of people affected by multiple BCCs developed BCCs on different sites during 7-years of follow-up.17 In our study, with 16-years of follow-up, 60% of people with multiple occurrences of BCC developed BCCs on more than 1 anatomic region, suggesting a common pathogenic mechanism is shared by BCCs across sites.
Regarding limitations of this study, although our findings did not suggest that sunburns contributed to BCC multiplicity, we could not exclude this possibility because of the known poor reproducibility of sunburn history.27 Consequently, we have not been able to assess if the acute intense pattern of UVR is associated with BCC multiplicity and site. Although some participants had a BCC before the observation period, the BCC records for that time were incomplete for some participants. We can therefore only be certain about the BCC status of all participants for the 16-year observation period. However, on repeating our analyses only in those who did not have previous BCC, similar results were obtained, and thus, we believe this has not influenced our findings to any material extent.
In conclusion, we have shown that among people affected by BCC, those who develop intermediate or high numbers of BCCs show evidence of severe sun damage both overall and for BCCs at different sites. How might our findings be useful as evidence for effective BCC prevention and healthcare? These findings suggest that control of BCC multiplicity in the population at large is feasible through decreasing cumulative sun exposure, thereby helping to curb the substantial encumbrance that BCC places on health systems serving white populations. Follow-up of BCC patients requires thorough whole-body skin examinations, rather than being limited to typically sun-exposed sites or to sites previously affected by BCCs.
The authors thank the Nambour Skin Cancer Study participants and their doctors who assisted in long-term monitoring of skin cancer. Many volunteer helpers from the Nambour community have given their invaluable support to our study. They thank Queensland Medical Laboratory and Sullivan Nicolaides Pathology for providing pathology reports for all study tumors. Naomi Richmond-Sinclair was supported by a Smart State PhD Scholarship, an ANZ Trustees Medical Research in Queensland PhD Scholarship and Queensland Institute of Medical Research PhD Top-Up Scholarship. Rachel Neale is supported by a NHMRC (Aust) Career Development Award.