Publ. Hlth. Lond. (1975) 90, 25-30
A Clinical Study of Skin Cancer: Genetic and Environmental Interactions B. A. Smithurst* M.B., B.S., M.P.H. (Harvard), F.R.A.C.P., M.F.CM., D.I.H. (Dundee)
H. Silverstonet M.A. (N.Z.), Ph.D. (Edinburgh)
Medical School, Herston Road, Herston, Queensland, Australia 4006 A survey of skin cancer patients attending the Queensland Radium Institute at the Royal Brisbane Hospital was carried out and the characteristics of the patients were compared with a control group of skin outpatients. The skin cancer patients had higher percentages o f patients o f Northern European ancestry and indoor occupations than the controls, although the skin reactions to the sun did not differ, which is at variance with other surveys. The diagnosis of cellular type, if made on clinical grounds only, was shown to have an 80~o chance of accuracy. The ratio of basal cell carcinomas to squamous cell carcinomas was 4.2 : 1 which is in line with the findings of others in this field.
Introduction It is a generally accepted belief that exposure to sunlight is a cause o f skin cancer. Evidence for this comes from observations by various researchers, that (a) pigmented races who burn much less easily than do white people rarely develop skin cancer (Urbach, 1964); (b) skin cancer occurs most often on exposed areas of head, neck, and arms and hands ( G o r d o n , Silverstone & Smithurst, 1972); (c) skin cancer in the white races is more c o m m o n in o u t d o o r than indoor workers (Silverstone & Searle, 1970); (d) skin cancer was most c o m m o n in fair-skinned, fair-complexioned a n d blue-eyed individuals (classically the Scotch-Irish type) (Silverstone & Searle, 1970); (e) prevalence rates for skin cancer rise with increases in latitude (Gordon et aL, 1972). In this pilot study an attempt is m a d e to compare environmental, racial and physical factors which m a y be of aetiological importance in skin cancer in a group o f patients demonstrating biopsy-proven basal and squamous cell c~/rcinomas (BCC and SCC) with an ageand sex-matched control group o.f patients who do not have any cutaneous malignancy. Methods of Survey Because o f the varying proportions o f BCC and SCC noted by previous workers, it was decided, with the co-operation o f the Queensland R a d i u m Institute, to carry out a survey of patients attending the R a d i u m Institute over a period o f some months in the early 1970s. One o f the authors attended the Queensland R a d i u m Institute Outpatients' Department at the Royal Brisbane Hospital to conduct the survey. *Reader in Social and Preventive Medicine, University of Queensland. tReader in Medical Statistics, Department of Social and Preventive Medicine, University of Queensland (Now deceased).
B. A. Smithurst and H. Silverstone
26
Patients were allotted to the survey in a random system. It was apparent that only a proportion o f the radiotherapists regularly saw minor malignancies and it was agreed that patients referred for interview from these radiotherapists would all have a biopsy of the lesion before the treatment was begun. A detailed questionnaire was administered to the patients by the researchers. A lesion site chart, based on the c h a r t produced by the New York University Medical Center's Oncology Section (Skin and Cancer Unit) was completed for each cancer case. These charts have the advantage of pinpointing the site of the lesion on an easily read numbered anatomical diagram. A total of 197 skin cancer patients was questioned and biopsied (patients with keratoses were excluded) and a similar questionnaire was administered to a matching control group of patients from the Skin Outpatients' Department who had no present or past history of skin cancer. The controls were matched for age and sex with the skin cancer patients; city dwellers predominated in the controls. As a result the controls also contained more individuals with indoor occupations than the cancer patients, since "household duties" were classified as an "indoor occupation". The various statistical analyses were adjusted to compensate for this deficiency in the matching.
Results of Survey Age and sex The control sample is 53.3 ~ male and the case sample is 6 4 - 0 ~ male; the control sample has 54-0~o in the 50-69 age group and the case sample has 63-5 ~ in the 50-69 age group. Therefore the control sample is slightly deficient in the male 50-69 age group, compared with the case sample. Table 1 shows the results o f matching cases and controls. Age is thought to be important in skin cancer epidemiology and because o f the age difference between the samples allowance was made in the statistical calculations by grouping ages into three categories, male 50, 50-64, and 65 and over. TABLEI. Controls and casesof skin cancers Cancer patients M F Total
M
20-29 30-39 ~--49 50-59 60-69 70-79 80-89
1 2 23 38 47 12 3
1 7 13 17 ~ 8 2
2 9 36 55 70 20 5
1 2 22 21 21 12 2
1 7 13 17 23 8 2
2 9 35 38 44 20 4
Total
126
71
197
81
71
152
Age
Controls F Total
Analysis o f chi-square was adopted for comparing the frequencies o f various chacteristics between the two groups. The simple chi-square test or a rank correlation test was used to assess associations between pairs of characteristics within each group. The difficulty in finding enough "controls" for the older male cancer patients is presumably a reflection of great susceptibilty of this group to skin cancer.
Ancestry of patiems Paternal and maternal ancestry were highly correlated within each group. However there was no significant difference noted between the cancer and the control group in relation to maternal and paternal ancestry (see Table 2).
A clinical study of skin cancer
27
Occupation Occupation was determined as " m a i n l y i n d o o r s " or " m a i n l y o u t d o o r s " (see T a b l e 3). T h e controls show a greater proportion o f individuals with indoor occupations than d o the cancer patients (68 % vs 56 %). T h e r e was n o t a significantly higher o u t d o o r exposure in the skin cancer group than in the controls. TABLE 2.
Cancer patients M F Total
M
Australian/Northern European Australian/Northern European +Scotch/Irish Scotch/Irish Other Unknown
102 I0
59 5
161 15
61 8
42 8
103 16
13 0 1
7 0 0
20 0 1
6 5 1
7 11 3
13 16 4
Total
!26
71
197
81
71
152
Patients' ancestry
Controls F Total
TABLE3. Occupation lndoo~ Outdoo~ Unknown Total
Cancer patients M F Total
M
Controls F Total
52 73 1
58 13 0
110 86 1
39 42 0
65 6 0
104 48 0
126
71
197
81
71
152
Complexion and eye colour A c o m p a r i s o n between the cases and controls showed that there were n o differences in complexion and eye colour between the two groups. As was to be expected, complexion and eye colour were closely correlated. T h e classification o f skin complexion was fair, m e d i u m and dark, and the eye colour, blue, grey or brown.
Sunburning The cases a n d controls were b o t h asked whether they had ever been severely sunburnt with blisters and whether their skin burnt easily in sunlight. T h e r e was n o difference in a history of severe sunburning between the cases and controls. A m o n g the skin cancer patients a history o f severe sunburning was significantly associated with an easily b u r n t skin (see Tables 4 and 5). TABLE4. Burn easily ?
Cancer patients
Controls
Yes
86 (43-6)
80 (52-0
No
lI l (56-4)
72 (47-4)
Total
197 (100-0)
152 (100.0)
B. A. Smithurst mid 11. Silverstone
28
Tanning N o difference could be detected between skin cancer patients and controls Jn relation to the degree of tanning (dark, light or absent) after sun exposure. TABLE 5.
Ever severely sunburned ?
Cancer patieuts
Yes No
94 (47-7) 103 (52.3)
76 (50.0) 76 (50.0)
Total
197 (100.0)
152 (100.0)
Controls
Hat wearing and outdoor sports Hat wearing in childhood was no more frequent in one g r o u p than in the other, nor were any differences revealed as far as o u t d o o r sporting activities were concerned. These findings in this small group have n o t been b o r n e out by some other studies in this field in which both going bareheaded in the sun and o u t d o o r sporting exposure to the sun have been significant factors in the pathenogenesis o f skin cancer.
Provisional clinical diagnosis and biopsy result Table 6 demonstrated the fact that even experienced clinicians m a y confuse BCCs with SCCs. There were 237 lesions which were diagnosed clinically as BCCs. One hundred and TABLE 6. Provisional clinical diagnosis Other malignancy
Keratosis benign
Total
BCC
SCC
Melanoma
B BCC i
153
6
0
5
2
166
SCC
17
23
0
0
1
41
2
1
0
3
0
6
0
0
0
0
0
O
3
5
0
0
1
9
62
O
0
0
0
67
237
35
0
8
4
289
0
P S
Melanoma
Y Other malignancy R Keratosis e benign s u
N o biopsy
l t
Total
seventy-five of these were biopsied and 153 were confirmed BCCs; 17 were SCCs and 5 were either melanomata or keratoses. This m e a n t tha t 13 ~o o f the BCCs diagnosed clinically as such were SCCs or melanomas or keratoses. A total of 261 lesions were seen b u t 62 of these were not biopsied for various reasons.
A clinical study o f skin cancer
29
Ratio o f BCCs to SCCs This gives a ratio of BCCs to SCCs o f 4"2 : 1 and is based on biopsy results (see Table 7). TABLE7. Age
BCCs
SCCs
20-29 30-39 40--49 50-59 60-69 70-79 80+
2 8 33 48 65 17 4
0 1 3 15 15 4 4
Total
177
42
Discussion
The selection of control patients for this project was made on the basis that, as the patients with skin cancer were hospital patients, a comparable sex and age matched group should also be taken from hospital patients--in this case from hospital dermatology out-patients. Skin cancer has been shown by past surveys (Gordon & Silverstone, in press) to have a higher incidence in older individuals o f Northern European and particularly Celtic (ScotchIrish) ancestry who burn easily on exposure to sun and who tan poorly, if at all. A classical person at risk is theindividual with a Scottish or Irish ancestry and probably name, with red hair, blue eyes, a pale skin and freckles. Long-term occupational exposure to sunlight also predisposes to skin cancer. The failure o f this study to reproduce some of the results o f previous epidemiologic studies in Australia could perhaps be explained by the fact that here a hospital population is being studied, a group which is notorious for overt and hidden epidemiologic bias and distorted correlations. It could have been that patients interviewed were aware of the conditions predisposing to skin cancer and because o f resentment at the questions or embarrassment at acquiring a preventable disease, they deliberately falsified their past history or did not reveal as much o f theirhistory as they could. Although this study was puzzling because of the apparent contradictions with other similar research, it did demonstrate the importance o f biopsy in making a diagnosis of skin cancer. There are skilled and experienced radiotherapists and dermatologists who can be reasonably confident in their diagnostic ability to distinguish between SCC and B C C b u t as these figures show, overall there is a 13 ~o error in a small series if clinical diagnosis is the sole criterion. It is known that skin histopathology is a specialized field o f pathology so it is instructive to observe that there was an 87 ~o chance of a correct diagnosis if the diagnosis was based on clinical grounds alone. Table 6 shows the chances that a P C C will be diagnosed as such clinically, and that an SCC will also be so diagnosed. It is apparent that the chances of clinical misdiagnosis o f an SCC are greater than the chances o f clinical misdiagnosis o f a BCC as 18 out of 23 (78.2~/o) were diagnosed most commonly as BCCs. The implications of this variation in diagnosis are important in both the treatment and prognosis of skin cancer and it is advisable that a combined approach to diagnosis b y employing biopsy for confirmation of clinical diagnosis be employed. In 67 cases studied, 62 of which were diagnosed as BCC, no biopsy was performed. The proportion of BCCs t o SCCs in this study based on biopsy diagnosis was 4-2 : 1 which correlates with similar figures elsewhere (Gordon et aL, 1972).
30
B. A. Smithurst and H. Silverstone
References Gordon, D. & Silverstone, H. (in press). World-wide Epidemiology o f Pre-malignant Cutaneous Lesions in Cancer of the Skin. Eds Andrade, R., Gumport, S. L., Popkin, G. L. & Rees, T. D. Philadelphia: W. E. Saunders & Co. Gordon, D., Silverstone, H. & Smithurst, B. A. (1972). The epidemiology of skin cancer in Australia. In Melanoma and Skin Cancer: Proceedings o f the International Cancer Conference Sydney 1972, 29. Ed. McCarthy, W. H. Australian Cancer Society and International Union against Cancer. Sydney: Government Printer. Silverstone, H. & Searle, J. H. G. (1970). The epidemiology of skin cancer in Queensland; the influence of phenotype and environment. British Journal of Cancer 24, 237. Urbach, F. (1964). Geographic pathology of skin cancer. In The Biologic Effects of Ultra-Violet Radiation. Pp. 635-50. Ed. Urbach, F. Oxford and New York: Pergamon Press.
A Clinical Study of Skin Cancer: Genetic and Environmental Interactions B. A. Smithurst* M.B., B.S., M.P.H. (Harvard), F.R.A.C.P., M.F.CM., D.I.H. (Dundee)
H. Silverstonet M.A. (N.Z.), Ph.D. (Edinburgh)
Medical School, Herston Road, Herston, Queensland, Australia 4006 A survey of skin cancer patients attending the Queensland Radium Institute at the Royal Brisbane Hospital was carried out and the characteristics of the patients were compared with a control group of skin outpatients. The skin cancer patients had higher percentages o f patients o f Northern European ancestry and indoor occupations than the controls, although the skin reactions to the sun did not differ, which is at variance with other surveys. The diagnosis of cellular type, if made on clinical grounds only, was shown to have an 80~o chance of accuracy. The ratio of basal cell carcinomas to squamous cell carcinomas was 4.2 : 1 which is in line with the findings of others in this field.
Introduction It is a generally accepted belief that exposure to sunlight is a cause o f skin cancer. Evidence for this comes from observations by various researchers, that (a) pigmented races who burn much less easily than do white people rarely develop skin cancer (Urbach, 1964); (b) skin cancer occurs most often on exposed areas of head, neck, and arms and hands ( G o r d o n , Silverstone & Smithurst, 1972); (c) skin cancer in the white races is more c o m m o n in o u t d o o r than indoor workers (Silverstone & Searle, 1970); (d) skin cancer was most c o m m o n in fair-skinned, fair-complexioned a n d blue-eyed individuals (classically the Scotch-Irish type) (Silverstone & Searle, 1970); (e) prevalence rates for skin cancer rise with increases in latitude (Gordon et aL, 1972). In this pilot study an attempt is m a d e to compare environmental, racial and physical factors which m a y be of aetiological importance in skin cancer in a group o f patients demonstrating biopsy-proven basal and squamous cell c~/rcinomas (BCC and SCC) with an ageand sex-matched control group o.f patients who do not have any cutaneous malignancy. Methods of Survey Because o f the varying proportions o f BCC and SCC noted by previous workers, it was decided, with the co-operation o f the Queensland R a d i u m Institute, to carry out a survey of patients attending the R a d i u m Institute over a period o f some months in the early 1970s. One o f the authors attended the Queensland R a d i u m Institute Outpatients' Department at the Royal Brisbane Hospital to conduct the survey. *Reader in Social and Preventive Medicine, University of Queensland. tReader in Medical Statistics, Department of Social and Preventive Medicine, University of Queensland (Now deceased).
B. A. Smithurst and H. Silverstone
26
Patients were allotted to the survey in a random system. It was apparent that only a proportion o f the radiotherapists regularly saw minor malignancies and it was agreed that patients referred for interview from these radiotherapists would all have a biopsy of the lesion before the treatment was begun. A detailed questionnaire was administered to the patients by the researchers. A lesion site chart, based on the c h a r t produced by the New York University Medical Center's Oncology Section (Skin and Cancer Unit) was completed for each cancer case. These charts have the advantage of pinpointing the site of the lesion on an easily read numbered anatomical diagram. A total of 197 skin cancer patients was questioned and biopsied (patients with keratoses were excluded) and a similar questionnaire was administered to a matching control group of patients from the Skin Outpatients' Department who had no present or past history of skin cancer. The controls were matched for age and sex with the skin cancer patients; city dwellers predominated in the controls. As a result the controls also contained more individuals with indoor occupations than the cancer patients, since "household duties" were classified as an "indoor occupation". The various statistical analyses were adjusted to compensate for this deficiency in the matching.
Results of Survey Age and sex The control sample is 53.3 ~ male and the case sample is 6 4 - 0 ~ male; the control sample has 54-0~o in the 50-69 age group and the case sample has 63-5 ~ in the 50-69 age group. Therefore the control sample is slightly deficient in the male 50-69 age group, compared with the case sample. Table 1 shows the results o f matching cases and controls. Age is thought to be important in skin cancer epidemiology and because o f the age difference between the samples allowance was made in the statistical calculations by grouping ages into three categories, male 50, 50-64, and 65 and over. TABLEI. Controls and casesof skin cancers Cancer patients M F Total
M
20-29 30-39 ~--49 50-59 60-69 70-79 80-89
1 2 23 38 47 12 3
1 7 13 17 ~ 8 2
2 9 36 55 70 20 5
1 2 22 21 21 12 2
1 7 13 17 23 8 2
2 9 35 38 44 20 4
Total
126
71
197
81
71
152
Age
Controls F Total
Analysis o f chi-square was adopted for comparing the frequencies o f various chacteristics between the two groups. The simple chi-square test or a rank correlation test was used to assess associations between pairs of characteristics within each group. The difficulty in finding enough "controls" for the older male cancer patients is presumably a reflection of great susceptibilty of this group to skin cancer.
Ancestry of patiems Paternal and maternal ancestry were highly correlated within each group. However there was no significant difference noted between the cancer and the control group in relation to maternal and paternal ancestry (see Table 2).
A clinical study of skin cancer
27
Occupation Occupation was determined as " m a i n l y i n d o o r s " or " m a i n l y o u t d o o r s " (see T a b l e 3). T h e controls show a greater proportion o f individuals with indoor occupations than d o the cancer patients (68 % vs 56 %). T h e r e was n o t a significantly higher o u t d o o r exposure in the skin cancer group than in the controls. TABLE 2.
Cancer patients M F Total
M
Australian/Northern European Australian/Northern European +Scotch/Irish Scotch/Irish Other Unknown
102 I0
59 5
161 15
61 8
42 8
103 16
13 0 1
7 0 0
20 0 1
6 5 1
7 11 3
13 16 4
Total
!26
71
197
81
71
152
Patients' ancestry
Controls F Total
TABLE3. Occupation lndoo~ Outdoo~ Unknown Total
Cancer patients M F Total
M
Controls F Total
52 73 1
58 13 0
110 86 1
39 42 0
65 6 0
104 48 0
126
71
197
81
71
152
Complexion and eye colour A c o m p a r i s o n between the cases and controls showed that there were n o differences in complexion and eye colour between the two groups. As was to be expected, complexion and eye colour were closely correlated. T h e classification o f skin complexion was fair, m e d i u m and dark, and the eye colour, blue, grey or brown.
Sunburning The cases a n d controls were b o t h asked whether they had ever been severely sunburnt with blisters and whether their skin burnt easily in sunlight. T h e r e was n o difference in a history of severe sunburning between the cases and controls. A m o n g the skin cancer patients a history o f severe sunburning was significantly associated with an easily b u r n t skin (see Tables 4 and 5). TABLE4. Burn easily ?
Cancer patients
Controls
Yes
86 (43-6)
80 (52-0
No
lI l (56-4)
72 (47-4)
Total
197 (100-0)
152 (100.0)
B. A. Smithurst mid 11. Silverstone
28
Tanning N o difference could be detected between skin cancer patients and controls Jn relation to the degree of tanning (dark, light or absent) after sun exposure. TABLE 5.
Ever severely sunburned ?
Cancer patieuts
Yes No
94 (47-7) 103 (52.3)
76 (50.0) 76 (50.0)
Total
197 (100.0)
152 (100.0)
Controls
Hat wearing and outdoor sports Hat wearing in childhood was no more frequent in one g r o u p than in the other, nor were any differences revealed as far as o u t d o o r sporting activities were concerned. These findings in this small group have n o t been b o r n e out by some other studies in this field in which both going bareheaded in the sun and o u t d o o r sporting exposure to the sun have been significant factors in the pathenogenesis o f skin cancer.
Provisional clinical diagnosis and biopsy result Table 6 demonstrated the fact that even experienced clinicians m a y confuse BCCs with SCCs. There were 237 lesions which were diagnosed clinically as BCCs. One hundred and TABLE 6. Provisional clinical diagnosis Other malignancy
Keratosis benign
Total
BCC
SCC
Melanoma
B BCC i
153
6
0
5
2
166
SCC
17
23
0
0
1
41
2
1
0
3
0
6
0
0
0
0
0
O
3
5
0
0
1
9
62
O
0
0
0
67
237
35
0
8
4
289
0
P S
Melanoma
Y Other malignancy R Keratosis e benign s u
N o biopsy
l t
Total
seventy-five of these were biopsied and 153 were confirmed BCCs; 17 were SCCs and 5 were either melanomata or keratoses. This m e a n t tha t 13 ~o o f the BCCs diagnosed clinically as such were SCCs or melanomas or keratoses. A total of 261 lesions were seen b u t 62 of these were not biopsied for various reasons.
A clinical study o f skin cancer
29
Ratio o f BCCs to SCCs This gives a ratio of BCCs to SCCs o f 4"2 : 1 and is based on biopsy results (see Table 7). TABLE7. Age
BCCs
SCCs
20-29 30-39 40--49 50-59 60-69 70-79 80+
2 8 33 48 65 17 4
0 1 3 15 15 4 4
Total
177
42
Discussion
The selection of control patients for this project was made on the basis that, as the patients with skin cancer were hospital patients, a comparable sex and age matched group should also be taken from hospital patients--in this case from hospital dermatology out-patients. Skin cancer has been shown by past surveys (Gordon & Silverstone, in press) to have a higher incidence in older individuals o f Northern European and particularly Celtic (ScotchIrish) ancestry who burn easily on exposure to sun and who tan poorly, if at all. A classical person at risk is theindividual with a Scottish or Irish ancestry and probably name, with red hair, blue eyes, a pale skin and freckles. Long-term occupational exposure to sunlight also predisposes to skin cancer. The failure o f this study to reproduce some of the results o f previous epidemiologic studies in Australia could perhaps be explained by the fact that here a hospital population is being studied, a group which is notorious for overt and hidden epidemiologic bias and distorted correlations. It could have been that patients interviewed were aware of the conditions predisposing to skin cancer and because o f resentment at the questions or embarrassment at acquiring a preventable disease, they deliberately falsified their past history or did not reveal as much o f theirhistory as they could. Although this study was puzzling because of the apparent contradictions with other similar research, it did demonstrate the importance o f biopsy in making a diagnosis of skin cancer. There are skilled and experienced radiotherapists and dermatologists who can be reasonably confident in their diagnostic ability to distinguish between SCC and B C C b u t as these figures show, overall there is a 13 ~o error in a small series if clinical diagnosis is the sole criterion. It is known that skin histopathology is a specialized field o f pathology so it is instructive to observe that there was an 87 ~o chance of a correct diagnosis if the diagnosis was based on clinical grounds alone. Table 6 shows the chances that a P C C will be diagnosed as such clinically, and that an SCC will also be so diagnosed. It is apparent that the chances of clinical misdiagnosis o f an SCC are greater than the chances o f clinical misdiagnosis o f a BCC as 18 out of 23 (78.2~/o) were diagnosed most commonly as BCCs. The implications of this variation in diagnosis are important in both the treatment and prognosis of skin cancer and it is advisable that a combined approach to diagnosis b y employing biopsy for confirmation of clinical diagnosis be employed. In 67 cases studied, 62 of which were diagnosed as BCC, no biopsy was performed. The proportion of BCCs t o SCCs in this study based on biopsy diagnosis was 4-2 : 1 which correlates with similar figures elsewhere (Gordon et aL, 1972).
30
B. A. Smithurst and H. Silverstone
References Gordon, D. & Silverstone, H. (in press). World-wide Epidemiology o f Pre-malignant Cutaneous Lesions in Cancer of the Skin. Eds Andrade, R., Gumport, S. L., Popkin, G. L. & Rees, T. D. Philadelphia: W. E. Saunders & Co. Gordon, D., Silverstone, H. & Smithurst, B. A. (1972). The epidemiology of skin cancer in Australia. In Melanoma and Skin Cancer: Proceedings o f the International Cancer Conference Sydney 1972, 29. Ed. McCarthy, W. H. Australian Cancer Society and International Union against Cancer. Sydney: Government Printer. Silverstone, H. & Searle, J. H. G. (1970). The epidemiology of skin cancer in Queensland; the influence of phenotype and environment. British Journal of Cancer 24, 237. Urbach, F. (1964). Geographic pathology of skin cancer. In The Biologic Effects of Ultra-Violet Radiation. Pp. 635-50. Ed. Urbach, F. Oxford and New York: Pergamon Press.
