This study investigated time trends and latitude differentials in the thickness distributions of invasive melanomas diagnosed in Australia between 1990 and 2006 using data from population-based cancer registries. Trends in incidence rates were calculated by sex, age group, thickness, year at diagnosis and latitude. For thin (<1.00mm) melanomas the increase was very pronounced during the early 1990s (1990–1996, annual percentage change and 95% confidence interval: males +5.6(+3.5,+7.7); females +4.1(+1.7,+6.5), but then incidence rates became stable among both males (+0.6(−0.1,+1.4)) and females (−0.0(−0.9,+0.9)) of all ages between 1996 and 2006. In contrast, incidence of thick (>4.00 mm) melanomas continued to increase over the entire period (males +2.6(+1.9,+3.4); females +1.6(+0.6,+2.6)). Recent reductions in the incidence of thin melanomas were observed among young (<50 years) males and females, contrasted by an increase in thin melanomas among older people, and increases in thick melanomas among most age groups for males and elderly (75+) females. A strong latitude gradient in incidence rates was observed, with rates being highest in northern, more tropical areas and lowest in the most southern regions. However, the magnitude of the increase in thick melanomas was most pronounced in southern parts of Australia. The observed trends in thin melanomas can most likely be attributed to the impact of early detection and skin awareness campaigns. However, these efforts have not impacted on the continued increase in the incidence of thick melanomas, although some increase may be due to earlier detection of metastasising melanomas. This highlights the need for continued vigilance in early detection processes.
With its high rates of ultraviolet radiation, outdoor lifestyle and predominately Caucasian population, Australia continues to have the highest incidence rates of cutaneous melanoma in the world, with an estimated age-standardised incidence rate (World 2000 population) of 40.2 cases per 100,000 population in 2008, similar to New Zealand (40.1).1, 2 Rates in other countries are substantially lower, with Switzerland (20.8), Denmark (19.9) and Norway (19.1) on the second tier of melanoma risk, and the USA (15.6), UK (11.6) and Germany (13.2) having lower incidence rates.1, 2 While melanoma has been the most rapidly increasing cancer among fair-skinned populations worldwide,3 studies from Europe, Canada, the United States and Australia have reported slowing or stabilizing rates of increase in incidence rates starting from the mid-1980s onwards.3–7 Much of the observed increase in melanoma incidence has been in thinner melanomas, with the incidence rates for thicker melanomas either increasing at a reduced rate or stabilizing.4, 6, 8, 9 With the strong association between tumour thickness and survival,8, 10–13 any increase in the incidence of thick melanomas has important implications for the mortality burden caused by melanoma.
Although other Australian studies have reported patterns of melanoma incidence over time by thickness4, 6, 8, 9 these have all been state-specific and have used varying analytical and reporting methods. To our knowledge, national trends on melanoma incidence by thickness have not been previously reported. In addition, the pooling of state-specific incidence data enables us to directly examine the association between incidence trends and geographical latitude, as has been recently reported for Norway.14
Material and Methods
All cases of invasive cutaneous melanoma (ICD-0-3 code C44, morphology codes M872-M879) diagnosed in Australia between 1990 and 2006 (inclusive) were obtained from the Australia Institute of Health and Welfare (AIHW) Australian Cancer Database (ACD). The ACD receives data from individual state and territory cancer registries on all cancers diagnosed for residents of Australia (except non-melanoma skin cancer). Notifications to these population-based registries are required by law. Information on sex, 5-year age group (up to 85+), calendar year of diagnosis and melanoma thickness (≤1.00 mm, 1.01–2.00 mm, 2.01–4.00 mm, >4.00 mm or Unknown) was provided. Throughout the article, thin melanomas refer to those < = 1 mm and thick melanomas are those >4 mm. The category ≤ 1.00 mm does not include in situ lesions. Cases with unknown age (n = 5) were excluded from all analyses.
To assess the impact of latitude on trends in melanoma thickness, we also obtained aggregated data from the AIHW on melanoma incidence for the eastern states and territories; Queensland, New South Wales, Australian Capital Territory, Victoria and Tasmania. Because of a smaller number of cases, data for Tasmania and Victoria were combined while the data for New South Wales included incident cases from the Australian Capital Territory. This provided 3 north-south geographical regions: “Northern” (latitudes ranging from ∼10° to 29° South), “Central” (from 28° to 37° South) and “Southern” (from 36° to 43° South). For these region-based data, age was collapsed by the AIHW prior to release into broad groups (<50 years, 50–64 years, 65–74 years and 75 years and over), as was period of diagnosis (1990–1996, 1997–2001, 2002–2006).
Directly age-standardised (World 2000) incidence rates were calculated according to region, sex, age group, thickness and year categories. We used Joinpoint software (version 3.4.3, National Cancer Institute, 2010)15 to assess the trends (measured by the annual percentage change, or APC), and specifically to determine whether there were any statistically significant changes in the magnitude or direction of the trends over the study period. The APC was estimated by fitting a regression line to the logarithm of the age-standardised rates with a linear term for year of diagnosis. To reduce the likelihood of reporting spurious changes in trends, we used a maximum of 3 joinpoints (i.e., up to 4 different trends) with a minimum of 5 years of data each.16 Monte Carlo permutation tests were used to examine the trend lines for each combination of join points, and the trend line that provided the best fit to the observed data was selected.17
Poisson regression models were used to investigate differences in melanoma incidence rates by region, sex, age group, thickness and year categories. The log of the estimated resident population,18 was used as the offset parameter in the Poisson model. Tests for interactions between the latitude regions and other statistically significant main effects were used to assess differences in effects by region. Separate models were used for thin and thick melanomas.
Approvals from the respective state and territory cancer registries were obtained by the AIHW before the data was released. No potentially identifying information was released outside the AIHW and therefore the AIHW Ethics Committee waived the requirement for Ethics Committee approval of this study.
Between 1990 and 2006, 139,943 Australian residents were diagnosed with an invasive melanoma, representing an average age-standardised (2000 World Population) incidence rate of 36.2 cases per 100,000 population (Table 1). Over half (56.2%) of the cases were males, and 67.2% of cases were diagnosed among people aged 50 years and over. Incidence rates were very low among children and younger adults (<12.2/100,000 up to the age of 24 years), then rapidly increased up to 173.5/100,000 for those aged 85 years and over. Over the whole study period, nearly two-thirds (62.7%) of invasive melanomas were thin, while 5.1% were thick. The proportion of melanomas that were thin increased from 58.0% in 1990 to a peak of 65.9% in 1999, and has ranged between 60.9% and 64.9% since then. In contrast, the proportion of melanomas that were thick gradually increased from 4.5% in 1990 to a peak of 5.9% in 2004 and 2006.
Table 1. Proportion and age-standardised rate (ASR) for incidence of invasive cutaneous melanoma classified by tumour thickness, Australia, 1990–2006
Trends by sex and thickness
Over the 17 years (1990–2006), incidence rates for melanoma generally increased within all known thickness categories (Table 2; Fig. 1). During the early to mid-1990s, incidence rates for thin melanomas increased at approximately twice the rate of increase of thick melanomas for both males and females (Table 2). However, in contrast with the generally ongoing increasing trends for intermediate (1.01–2.00 mm and 2.01–4.00 mm) and thick melanomas, the Joinpoint analysis suggested that the increasing trends for thin melanomas by sex plateaued (non-significant trend) during the period 1996–2006 (Table 2).
Table 2. Annual percentage change (APC) in age standardised incidence rates for invasive melanoma by trend period, Australia, 1990–2006
Trends by age group and thin/thick melanomas
Until the mid-1990s significantly increasing trends in thin melanomas were observed across all 4 age groups for males and females (Table 3). The gradients of the trend generally increased with age among males and were much steeper than the corresponding trends among females aged 50 years and older. However from 1997 onwards, there was a significant decrease in the incidence of thin melanomas among younger (<50 years) males and females. The magnitude of the rate of increase also reduced for males in each of the other age groups from around the mid-1990s.
Table 3. Annual percentage change (APC) in age standardised incidence rates for thin- and thick-invasive melanomas by sex, age at diagnosis and trend period, Australia, 1990–2006
While there was some evidence of consistently increasing trends for thick melanomas across all age groups, for some of the age-sex cohorts (males aged 50–64 years and females aged <75 years) these trends were not statistically significant (Table 3).
Incidence by latitude
There was a strong inverse relationship in melanoma incidence by latitude, with incidence rates between 1990 and 2006 being highest in the Northern region which is closest to the equator (average of 52.7/100,000), then lower in the Central and Southern regions (Central: 37.5/100,000; Southern: 30.9/100,000). This pattern was also true for each time period (Table 4). Compared with melanoma incidence rates in the Northern region, rates were 28% lower in the Central region and 43% lower in the Southern region after adjustment for sex, age group, diagnosis year and thickness (Table 5). The corresponding rates for thin melanomas by region followed a similar, although even more pronounced, pattern. In contrast, the incidence rate of thick melanomas in the Central region was equivalent to the Northern region, while the rate for the Southern region was 41% lower.
Table 4. Differences in age-standardised incidence rates of melanoma classified by tumour thickness and latitude, Australia, 1990–2006
Table 5. Incidence rate ratios (with 95% confidence intervals) of melanoma by region, sex, age group, year and thickness
As evidenced by the significant (p < 0.01) tests for interaction between region and time, the trends over time were significantly different across the 3 regions for all melanomas and both thin and thick melanomas. For example, compared with the region-specific incidence of thick melanoma between 1990 and 1996, there was a 19% increase in the Northern region by 2002 and 2006, while the Central region increased by 31%, and the Southern region increased by 63% over the same period (Table 5).
This large study, utilising data from almost 140,000 Australians diagnosed with melanoma, demonstrated that the previously sharp rate of increase in thin melanomas has plateaued. However, rates of thick melanoma have continued to increase among both males and females, particularly in areas of higher latitude.
In contrast to these national results, a recent Queensland study4 examining trends between 1991 and 2002 found no evidence to suggest that the increasing trends for thin melanomas were levelling off. Similarly, ongoing increasing trends have been reported previously in the United Kingdom (1993–2003),8, 19 the United States (1988–1994),20 New South Wales (1989–1996)6 and (1993–2003),8 Puerto Rico (1987–2002),21 Southern Germany (1976–2003),22 and Northern Ireland (1984–2006).23 However the short time periods and analytical methods used in some of these studies may not have allowed changes in trends to be detected.
It has been suggested that widespread increases in thin melanomas during the 1990s were predominately due to heightened levels of melanoma awareness and improved early detection, rather than a real increase in the underlying melanoma incidence.4, 24 The recent levelling off of this increase observed in the Australian context is consistent with this hypothesis—improved detection leads to an initial large increase in the incidence of early disease after which incidence rates plateau, albeit at a higher level than previously, as the pool of undetected lesions diminishes.25, 26 It remains to be seen whether this stabilization in the incidence of thin melanomas continues, or even whether incidence trends will eventually start to decrease across the whole population.
To the extent that the apparent increased detection of thin melanomas observed here is explained by earlier diagnosis (as opposed to over-diagnosis of non-progressive lesions), it would be expected to be followed in time by a reduced incidence of thick melanomas. These Australian results suggest the opposite has occurred. That is, the improvements in early diagnostic methods have not been sufficient to counter an increase in the incidence in the underlying incidence of thicker tumours. One explanation is that some tumours grow rapidly and are thus less likely to be detected before they become thick. Therefore advances in early detection activities such as skin screening may be less likely to have an impact on the incidence rates of these types of tumours.27
Increased tumour thickness has been consistently shown to be the strongest predictor of poorer survival prognosis, both internationally8, 10–13 and in Australia.28–30 When combined with the limited effectiveness of treatment for thick melanomas,31 it would be expected that any trends in thick melanomas, after allowing for some lead time, would be reflected in melanoma mortality outcomes. Therefore, the consistently increasing incidence of thick melanomas observed in this study should be accompanied by increasing mortality rates of melanoma. This has not been observed. A previous study reporting on Australian melanoma mortality rates since 1950, found that, following consistent increases, melanoma mortality rates stabilized among males between 1989 and 2002 and decreased among females.31 Additional analyses (results not shown) suggest these patterns have continued up to 2006.
One possible explanation is that trends in the proportion (9%) of melanomas with unknown thickness, nearly twice as common as thick melanomas (5%), could compromise the reported trends in thick melanomas, particularly since the percentage of melanomas with unknown thickness has decreased over time, from 12% in 1990 to 8% in 2006. Reductions in the percentage of melanomas with unknown thickness have also been reported in the United States.24 Although we were unable to obtain national data, unpublished data from the Queensland Cancer Registry showed that over well over half (60%) of the melanomas with unknown thickness were diagnosed on the basis of histology of metastasis. If previously metastasising melanomas are now being detected earlier, most likely as thick melanomas, then this would increase the number of thick melanomas being detected, and hence directly impact on observed incidence trends by thickness. However the relationship between thick melanomas and melanomas with unknown thickness is not consistent by latitude; the rates of thick melanomas have increased in each of the 3 regions whereas the rates of melanomas with unknown thickness has either decreased (Northern), remained stable (Central) or increased (Southern). Also the trends in melanomas of unknown thickness are highly variable, suggesting against a consistent improvement in detection practices. Therefore the possible implications of changes in the rates of melanomas with unknown thickness on the observed incidence trends by thickness can only be speculative. Although some of the observed increase in thick melanomas may be due to improved detection of previously metastasising melanomas, it would be presumptuous to suggest that a real increase has not occurred. Clearly further investigation is required.
Our finding that the incidence of thin melanoma is decreasing among younger people is promising. The importance of early life sun exposure and the generally long latency period for the development of melanoma is well established.32 Public primary prevention campaigns directed at reducing sun exposure have been ongoing in Australia since the early 1980s.4, 33 The observed reduction in thin melanomas in younger age groups, and a similar stabilising of rates of non-melanoma skin cancer,34 could lend cautious support to the success of these campaigns, particularly if these lower rates continue into the future.4, 34, 35
As suggested by others,24 the current analysis highlights a need for new early detection strategies to be developed, particularly for segments of the population previously shown to be at higher risk of thick melanomas, including men, older people and those with low education.36 These strategies should include identification of the clinical features of melanomas that are more likely to have a rapid growth phase, and encouraging regular clinical skin examinations among high risk individuals. A recent case-control study of melanoma screening demonstrated that having a whole-body clinical skin examination in the 3 years prior to a diagnosis of melanoma was associated with a 40% reduction in the incidence of melanoma ≥3.00mm thick.37 Therefore, more research is needed to understand differences in the development between thin and thick melanoma which would then allow specific targeting of the various underlying factors in future public health campaigns.
We found a strong association between melanoma incidence and latitude, with overall incidence risks being significantly lower in the Central and Southern regions compared with the Northern region. A similar pattern was observed for thin melanomas, but for thick melanomas there was only a significant differential between the Southern and Northern regions. We also found strong evidence that the changes in incidence over time varied by latitude, with the increase in incidence of thick melanomas over time particularly pronounced in Southern states. Reasons for this differential increase are unclear. The authors of a European study38 suggested possible reasons for an observed differential in the incidence of thick melanoma by latitude were variations in overall awareness of melanoma and frequency of campaigns aimed at early detection. However, it is unlikely that factors such as lower prevalence of melanoma, impacting on clinical diagnostic abilities,39 or lower community awareness of melanoma, are explanations for the observed Australian trends. Per capita expenditure on sun protection programs is reportedly much higher in Victoria (Southern region) than Queensland (Northern region),40 however the increased incidence of non-melanoma skin cancer in Northern Australia34 may indirectly increase awareness through greater utilisation of doctors for treatment.
The strengths of this study include the use of data from population-based cancer registries providing a complete enumeration of all Australians diagnosed with invasive melanoma between 1990 and 2006, including information about melanoma thickness at diagnosis. All information used in this study has been collected prospectively for administrative purposes independently of the study hypotheses, thus removing recall or information bias. In contrast to other research studies that have considered constant linear changes over their entire study period, the use of Joinpoint regression enabled us to detect changes in the magnitude and direction of trends over time. Since the selection of joinpoints can be influenced by random fluctuations in rates, particularly those at the end points, we chose conservative parameters to reduce the chance of detecting spurious changes in trends. We were unable to obtain region-specific data by calendar year due to confidentiality restrictions, so the aggregating of incidence data across year groups may have influenced our ability to recognise trends over time by latitude.
There are encouraging trends in the incidence of thin melanomas; however, the continued increase in the incidence of thick melanomas, more pronounced in regions of high latitude, is cause for concern. Although improvements in early detection could mean that some melanomas that may previously have metastasised are now being diagnosed as thick melanomas, there is still a strong need to develop new early detection strategies, ideally with an emphasis on those groups known to have greater risk of being diagnosed with thick melanomas, to improve the efficacy of detecting melanoma early and address this increasing trend of thick melanomas.
The authors acknowledge the assistance of staff from the Cancer and Screening Unit, Australian Institute of Health and Welfare who conducted the data extraction required for this study.
Dr. Peter Baade is supported by the National Health and Medical Research Council (CDF 1005334).
The authors appreciate the assistance provided by Jacques Ferlay, International Agency for Research on Cancer (IARC) who re-calculated the international age standardized melanoma incidence rates from GLOBOCAN using the World 2000 Standard population as described in the Introduction.