Trends in prognostic factors and survival from cutaneous melanoma in Yorkshire, UK and New South Wales, Australia between 1993 and 2003
Article first published online: 23 MAY 2008
Copyright © 2008 Wiley-Liss, Inc.
International Journal of Cancer
Volume 123, Issue 4, pages 861–866, 15 August 2008
How to Cite
Downing, A., Yu, X. Q., Newton-Bishop, J. and Forman, D. (2008), Trends in prognostic factors and survival from cutaneous melanoma in Yorkshire, UK and New South Wales, Australia between 1993 and 2003. Int. J. Cancer, 123: 861–866. doi: 10.1002/ijc.23495
- Issue published online: 12 JUN 2008
- Article first published online: 23 MAY 2008
- Manuscript Accepted: 16 JAN 2008
- Manuscript Received: 13 NOV 2007
The aim of this study was to compare trends in prognostic factors and survival from cutaneous melanoma between 1993 and 2003 in 2 populations with dramatically different underlying incidence rates [Yorkshire, UK, and New South Wales (NSW), Australia] and to look at whether the greater investment in melanoma prevention and early detection in Australia has resulted in any relative differences in survival. Patients diagnosed with invasive melanoma between 1993 and 2003 in Yorkshire (n = 4,170) and NSW (n = 30,520) were identified from cancer registry databases and prognostic information (age, sex, socioeconomic background, tumour site and Breslow thickness) was extracted. Age-standardised incidence rates, 5-year relative survival and relative excess risk of death were calculated. Between 1993–1995 and 2001–2003, the incidence of melanoma increased in both areas. These increases were mainly seen in tumours with thickness ≤1 mm. Five-year relative survival was 86.9% (95% CI 85.2–88.5) in Yorkshire and 88.6% (95% CI 88.1–89.1) in NSW. Compared with that in NSW, survival in Yorkshire was lower for males and for those living in the most deprived areas. Despite the increase in good prognosis of thin tumours, there was no significant change in survival over the time period in either area. After adjustment for differences in prognostic factors, the relative excess risk of death in Yorkshire compared to that in NSW reduced from 1.36 (95% CI 1.20–1.53) to 1.11 (95% CI 0.99–1.23). Differences in tumour thickness appeared to be the most important factor. © 2008 Wiley-Liss, Inc.
Recent studies of populations in Europe show that the incidence of cutaneous melanoma (CM) is continuing to increase across all age groups.1–3 By comparison, studies in Australia and North America suggest that incidence has stabilised, or even fallen, mainly in the younger age groups.4–6 At the same time, the proportion of thin tumours has been increasing.3, 5, 7 Breslow thickness, the depth of the tumour from the granular layer of the epidermis, is the most important prognostic factor in survival from CM.8, 9 Thin tumours have a very good prognosis, and surgical excision is often curative, but for thick or ulcerated tumours, which demonstrate an increased risk of metastatic disease, the prognosis is less encouraging.10 Public health campaigns highlighting the risks associated with sun exposure have been in place in Australia since the 1960s and in the UK since the 1980s, and the recent trends seen in CM incidence (and survival) have been attributed, at least in part, to these campaigns.
The aim of this study was to compare trends in prognostic factors and survival from CM between 1993 and 2003 in 2 populations with dramatically different underlying incidence rates [Yorkshire, UK, and New South Wales (NSW), Australia] and to look at whether the greater investment in melanoma prevention and early detection in Australia has resulted in any relative differences in survival.
Material and methods
Cases of primary invasive melanoma (ICD1011 code C43) diagnosed between January 1, 1993 and December 31, 2003 were identified from 2 data sources: the Northern and Yorkshire Cancer Registry and Information Service (covering the former Yorkshire Health Authority, UK) and the NSW Central Cancer Registry (covering the state of NSW, Australia). Relevant patient and tumour characteristics (age, sex, Breslow thickness, histological type, and body site) were extracted. Tumour Breslow thickness was grouped as follows: ≤1.0 mm, 1.01–2.00 mm, 2.01–4.0 mm and >4.0 mm, according to the American Joint Committee on Cancer staging criteria.12 A thickness measurement was missing in 381 (9.1%) of the tumours in Yorkshire and 3,154 (10.3%) of those in NSW. An area-based measure of socioeconomic background (SEB) was added using the postcode of residence at the time of diagnosis (for Yorkshire) and Local Government Area (for NSW) and divided into quintiles. For the Yorkshire data, the 1991 Townsend Index was used.13 For the NSW data, the 1996 Index of Relative Socioeconomic Disadvantage was used.14
World age-standardised incidence rates were calculated as 3-year rolling averages using the 1998 population from each area.15 Patients were followed up to December 31, 2004 for survival status. The survival time was measured as the time difference between the date of diagnosis and the date of death or the date of censoring. The relative survival method was used in this study.16 In addition, relative excess risks of death (RERs) due to diagnosis of melanoma were calculated using Poisson regression models.17, 18 Ninety-five percent confidence intervals (CIs) for the RERs were calculated using the estimated coefficients and standard errors from the Poisson models. To estimate how much of the survival difference was due to SEB, tumour site and thickness, 4 models were fitted.19 The first model was adjusted for age group at diagnosis and sex; the second to the fourth included additional adjustment for SEB, tumour site and thickness respectively. All analyses were performed using STATA 9.0 (StataCorp, TX) and SAS 8.2 (SAS Institute, NC).
The world age standardised rate of CM in Yorkshire rose from 4.9 per 100,000 in males and 7.5 per 100,000 in females in the 3-year period 1993–1995 to 7.9 per 100,000 and 10.7 per 100,000 in males and females in 2001–2003. In NSW, incidence was higher in males than in females. The rate increased from 23.8 per 100,000 in females and 33.3 per 100,000 in males to 29.5 per 100,000 and 43.7 per 100,000 in females and males during the same time period.
In Yorkshire, the increase in incidence was driven mainly by an increase in thin tumours (≤1.0 mm), which rose from 1.8 to 4.0 per 100,000 in males and from 4.0 to 6.5 per 100,000 in females over the study period (Figs. 1a and 1b). There were also increases in the other thickness groups. In NSW, the rate of thin tumours increased over the study period but at a slower rate than in Yorkshire, from 19.3 to 24.8 per 100,000 in males and from 15.3 to 18.6 per 100,000 in females. Increases in the other thickness groups were much smaller, and in females the incidence of thick tumours appears to have stabilised.
The trends in the incidence of CM by age group were similar in both areas, with the highest rates in the older age groups (over 65 years) (Figs. 2a and 2b). By 2001-2003, the rates in those over 65 years reached 29.0 per 100,000 in Yorkshire and 168.6 per 100,000 in NSW. The main difference between the 2 areas was the stable rate in those aged 15–44 years in NSW compared to an increase in recent years in Yorkshire. No differences by sex were observed.
The incidence of invasive superficial spreading melanoma increased from 3.4 to 6.3 per 100,000 in Yorkshire and from 12.2 to 15.8 per 100,000 in NSW over the study period. (Table I) There was also a smaller increase in the incidence of nodular melanoma in NSW. However, the many trends are difficult to interpret due to the high proportion of tumours recorded as unspecified histological type (15.1% in Yorkshire and 34.6% in NSW).
|New South Wales|
The pattern of incidence by body site was similar for Yorkshire and NSW, with melanomas of the trunk being most common in males and tumours of the lower limb most common in females (Figs. 3a and 3b). However, in males, tumours of the trunk were relatively more common in NSW than in Yorkshire and, in females, tumours of the lower limb were relatively more common in Yorkshire.
The relationship between incidence and SEB was stronger in Yorkshire, with higher incidence rates in those living in the most affluent areas and lower rates in those living in the most deprived areas (Figs. 4a and 4b). The difference in rates between the most affluent and most deprived quintiles also increased over the study period. These patterns were not seen in NSW, where the highest rates were seen in quintile I (the most affluent) followed by quintiles IV and V (the most deprived).
Five-year relative survival in Yorkshire was 86.9% (95% CI 85.2–88.5) and in NSW this was 88.6% (95% CI 88.1–89.1). (Table II) Females had greater survival rates than males in both areas, but the survival gap was considerably larger in Yorkshire. Survival in Yorkshire was highest for patients with tumours of the lower limb, with tumours of the trunk having the worst prognosis. In NSW, survival was highest for patients with tumours of the trunk and lower limb and lowest in those with head and neck tumours. The prognosis was better for patients diagnosed with the thinnest tumours (99.0%; 95% CI 98.5–99.4 in NSW and 101.1%; 95% CI 90.4–106.2 in Yorkshire) and was worse for those diagnosed with the thickest tumours (54.5%; 95% CI 51.1–57.8 in NSW and 53.9%; 95% CI 45.2–62.3 in Yorkshire). Analysis by period of diagnosis showed no significant change in survival over time in either area. The analysis was repeated excluding the thin tumours (≤1.0 mm) and the overall 5-year survival reduced to 73.1% (95% CI 70.3–75.8) in Yorkshire and 73.0% (95% CI 72.0–74.0) in NSW. Again, there was no change in survival over time in either area.
|Yorkshire||New South Wales|
|No. cases||RSR (%)||95% CI||No. cases||RSR (%)||95% CI|
|I (most affluent)||1,242||91.9||(87.9–95.3)||6,657||90.4||(89.3–91.4)|
|V (most deprived)||317||75.9||(66.2–84.2)||6,187||88.0||(85.4–90.4)|
|Head and neck||742||85.8||(78.5–92.4)||5,147||83.0||(81.5–84.4)|
Before adjusting for other prognostic factors, the relative excess risk of death in Yorkshire compared to NSW was 1.36 (95% CI 1.20–1.53), meaning that patients in Yorkshire had 36% higher excess mortality due to diagnosis of melanoma compared with patients in NSW. (Table III) After additional adjustment for SEB, the RER increased to 1.40 (95% CI 1.24–1.58). Further adjustment for tumour site reduced the risk to 1.22 (95% CI 1.08–1.39). In the final model, after additional adjustment for Breslow thickness, the difference in survival between the 2 populations reduced to a borderline nonsignificant level with the RER being 1.11 (95% CI 0.99–1.23).
|Model 11||Model 21||Model 31||Model 41|
|RER||95% CI||RER||95% CI||RER||95% CI||RER||95% CI|
|New South Wales||1.00||1.00||1.00||1.00|
|I (most affluent)||1.00||1.00||1.00|
|V (most deprived)||1.36||(1.19–1.55)||1.32||(1.16–1.51)||1.08||(0.96–1.22)|
|Head and neck||1.63||(1.45–1.85)||0.99||(0.89–1.11)|
In addition to the trends in invasive melanoma, we have also analysed the trends in in situ tumours recorded by the 2 registries (1,335 in Yorkshire and 11,995 in NSW). During the study period, there was a large increase in the incidence of in situ melanoma in NSW (Fig. 5). The rate increased from 7.3 and 6.0 per 100,000 in males and females in 1993–1995 to 22.9 and 16.4 per 100,000 in 2001–2003. The largest increase was seen for lentigo maligna, which increased from 1.5 to 6.8 per 100,000 (males and females combined). In Yorkshire, the rate increased only slightly from 1.4 to 1.7 per 100,000 in males and from 2.4 to 2.6 per 100,000 in females during the study period. Superficial spreading melanoma increased the most from a rate of 0.3 to 1.0 per 100,000.
Over the period of 1993–2003, there were increases in the incidence of CM in Yorkshire and in NSW. Most of the increase in incidence has been driven by an increase in thinner tumours. However, the increase in thin tumours has not translated into an improvement in survival in either area during the period of this study. Whilst the proportion of thinner tumours has increased, the incidence rates of intermediate and thick tumours have continued to increase slowly (apart from thick tumours in females in NSW). Even though there are substantially more thin tumours over time, the survival rate for these is so good (∼100%) that it makes little difference to the overall survival rate. Survival is now almost entirely being driven by the intermediate and thick tumours, and this has not substantially changed during the period of this study. Additional analysis of mortality data over the same period showed no significant annual percentage change in mortality in either area (results not shown) indicating that control of this poor prognosis sub-group will be necessary to improve outcomes.
Before adjusting for prognostic factors, the risk of death was higher in Yorkshire than in NSW. Tumour thickness appeared to account for the majority of this excess risk, as the RER reduced to borderline/non-significance after adjustment. This may be an indication of success in educating the public about the risks of sun exposure in Australia. Campaigns highlighting these risks began in Australia in the 1960s but in other parts of the world, including the UK, prevention programmes were started much later. It has been shown that such programmes can influence sun-related behaviours.20 However, due to the lag between exposure and the development of melanoma (most common in the elderly), it may be many years before any benefits of primary prevention campaigns are seen. Health education usually includes both sun avoidance or protection (primary prevention) and skin awareness and early consultation (secondary prevention). Early detection of melanoma may lead to a higher proportion of thinner tumours being diagnosed, but it is unlikely to have any substantial impact on melanoma mortality. It may prove much more difficult to improve survival further. One reason for the failure to promote earlier diagnosis of thicker tumours may be relative resistance to health promotion campaigns in a proportion of the population. A second reason may be the presence of a tumour that is biologically aggressive from the beginning and grows very quickly, for which early detection is a much more difficult issue.
The incidence of CM was higher in males than in females in NSW, but the opposite was seen in Yorkshire. Survival, however, was lower in males in both areas, although the differential was less in NSW. This does not appear to be due to differences in stage at diagnosis between males and females.3 The trends observed in melanoma incidence by body site are similar to those reported previously, with tumours of the trunk being most common in males and lower limb tumours being most common in females.1, 4 Incidence has increased both in the intermittently exposed sites (trunk and leg) and the chronically sun-exposed sites (head and neck and arm) in Yorkshire and NSW. Melanomas of the head and neck or arm are more likely to result from a lifetime of sun exposure, which may be difficult to halt (especially in the sunnier climate of NSW). However, melanomas of the intermittently exposed sites (held to be related etiologically to recreational sun exposure21) are still increasing, despite the public health campaigns. In NSW, survival was highest for tumours of the trunk (along with tumours of the leg). This finding appears to contradict previous results, where tumours of the trunk are associated with a worse prognosis22, 23 as was seen in Yorkshire in this study. We have no explanation for this.
Another difference between the populations was the relationship of both incidence and survival with SEB. This was much stronger in Yorkshire, with the more affluent having a higher incidence, but in contrast, the more affluent have a better prognosis.3 The higher incidence in more affluent people within the UK has been attributed to greater access to holidays in sunnier climates, but this effect may diminish over time as cheaper air fares increase mobility. In Australia, where the whole population has very high sun exposure, there were no clear differences in incidence according to social status. Whilst some differences in survival by SEB have been observed in Australia, these are small and not consistent across the socioeconomic scale.22, 24
According to these data, there has been a substantial increase in the number of in situ melanomas diagnosed in NSW over the period of this study. Similar trends have been observed in Queensland, Australia,5 Sweden,25 and the US.26 In contrast, in Yorkshire, the incidence of in situ tumours remained relatively stable. The recording of in situ tumours is less reliable than that for invasive tumours and the increase may simply represent an increase in registration. Alternatively, the increases could represent an “overdiagnosis” due to increased scrutiny and an increased number of biopsies being taken.27 The largest increase was seen for lentigo maligna, which is related etiologically to chronic sun exposure and most commonly diagnosed in the elderly. Correspondingly, the largest increases in incidence were seen in those aged over 65. In contrast, in Yorkshire, the incidence of in situ tumours remained relatively stable.
There are several limitations associated with this study. Firstly, the life tables used to calculate relative survival in NSW were based on more recent years compared to those for Yorkshire, and this could have led to an overestimate of survival in Yorkshire and, therefore, an underestimate of the difference in survival between the 2 populations. Secondly, a large proportion of the histological subtypes were recorded as unspecified (15% in Yorkshire and 35% in NSW), making the trends in incidence for these subtypes difficult to interpret. In addition, Breslow thickness was missing in 9% and 10% of cases in Yorkshire and NSW, respectively, but we do not feel this will have had any substantial effect on the results.
This study has shown that the incidence of CM continues to rise in both Yorkshire and NSW. This is mainly due to an increase in the diagnosis of thin tumours, but there has been no decrease in the incidence of thicker tumours. As a result, survival has not increased over the period of this study in either area. There is an excess risk of mortality in Yorkshire compared with that in NSW, and this is mainly explained by differences in tumour thickness. Finally, there are indications that incidence rates are stabilising in the younger age groups in NSW. This may be due to stabilisation of the incidence of tumours on intermittently sun-exposed parts of the body. Any changes in chronically exposed sites will probably take longer to be seen, and as life expectancy generally increases may be more difficult to modify.
We thank Professor Bruce Armstrong for his comments on the draft manuscript and staff at the NSW Central Cancer Registry for supplying the data.
- 1Cancer Research UK. UK Malignant Melanoma incidence statistics. 2005. [Internet] 2005 [Cited 2007 November]. Available from http://info.cancerresearchuk.org/cancerstats/types/melanoma/incidence/.
- 11World Health Organisation. International statistical classification of diseases and related health problems 10th revision. Geneva: WHO, 1992.
- 12American Joint Committee on Cancer. Manual for staging of cancer. Philadelphia: Lippincott, 2002. 209–20.
- 13Health and deprivation. Inequality and the north. London: Routledge, 1988., .
- 14Australian Bureau of Statistics. Information paper: 1996 Census Socio-Economic Indexes For Areas (SEIFA). Canberra: ABS, 1998.
- 15Cancer mortality for selected sites in 24 countries (1950–57). Sendai, Japan: Department of Public Health, Tohoku University of Medicine, 1960..
- 23Malignant tumors (melanomas and related lesions). Atlas of tumor pathology: melanocytic tumors of the skin 2 (Third series). Washington, DC: AFIP (Armed Forces Institute of Pathology); 1991: 103–205., .