An ecologic study of cancer mortality rates in Spain with respect to indices of solar UVB irradiance and smoking

Abstract

There is increasing evidence that vitamin D reduces the risk of many types of cancer. Geographic variations in cancer mortality rates in Spain are apparently linked to variations in solar ultraviolet (UV) irradiances and other factors. Cancer mortality rates for 48 continental Spanish provinces for 1978–1992 were used in linear regression analyses with respect to mortality rates for latitude (an index of solar UVB levels), skin cancer (an index of high cumulative UVB irradiance), melanoma (an index related to solar UV irradiance and several other factors) and lung cancer (an index of cumulative effects of smoking). The 9 cancers with mortality rates significantly correlated with latitude for 1 or both sexes were brain, gastric, melanoma, nonmelanoma skin cancer (NMSC), non-Hodgkin's lymphoma (NHL), pancreatic, pleural, rectal and thyroid cancer. Inverse correlations with latitude were found for laryngeal, lung and uterine corpus cancer. The 17 cancers inversely correlated with NMSC are bladder, brain, breast, colon, esophageal, gallbladder, Hodgkin's lymphoma, lung, melanoma, multiple myeloma, NHL, ovarian, pancreatic, pleural, rectal, thyroid and uterine corpus cancer. The 16 correlated with melanoma are bladder, brain, breast, colon, gallbladder, leukemia, lung, multiple myeloma, NHL, ovarian, pancreatic, pleural, prostate, rectal, renal and uterine corpus cancer. The results for lung cancer were in accordance with the literature. These results provide more support for the UVB/vitamin D/cancer hypothesis and indicate a new way to investigate the role of solar UV irradiance on cancer risk. They also provide more evidence that melanoma and NMSC have different etiologies. © 2006 Wiley-Liss, Inc.

Evidence continues to mount that vitamin D produced from solar ultraviolet B (UVB) irradiance or taken orally reduces the incidence and mortality rates of and increases the survival for many forms of cancer. The UVB/vitamin D/cancer hypothesis was originally proposed in 1980 with respect to colon cancer1 and has now been extended to about 20 types of cancer2, 3, 4 including melanoma5, 6 and non-Hodgkin's lymphoma.7, 8 Although much of the research linking vitamin D to cancer risk reduction is observational, there is sufficient supporting evidence for mechanisms,9, 10 dose-response relationships,3, 11 replications in diverse populations12, 13 and seasonal variations in cancer survival rates14, 15, 16, 17, 18 that the criteria for causality in a biological system19 are generally satisfied if not fully accepted. One of the objections is the use of latitude as an index of solar UVB irradiance and vitamin D production in ecologic studies.

Spain has an excellent mortality rate dataset for 30 types of cancer for 52 provinces, spanning 1978–1992.20 Such data would be useful in trying to learn more about the causes of cancer21 and have been used in previous studies to investigate the geographic variation in cancer mortality rates with respect to such factors as income level, rural residence and average parity,22 socioeconomic level and farming,23 and mining and industry.24

Spain is a Mediterranean country lying between the latitudes of 36° and 44° N. Solar UVB is the primary source of vitamin D for the inhabitants, since dietary sources are primarily fish, which does not provide much vitamin D for elderly Spanish,25 and supplements are not regularly consumed.25, 26, 27, 28 On the basis of the UVB doses for the eastern United States for July 1992, the daily doses should range from 6.5 kJ/m² in the south to about 4.0–4.5 kJ/m² in the north in summer. However, the amount of time people spend in the sun is often related to occupation and recreational activities as well as ambient temperature. Since Spain is populated largely by those with strong ancestral ties to the country, the skin pigmentation is largely adapted to the solar UV in Spain.28 Thus, Spain is a country where the amount of UVB should affect the risk of cancer, but developing a vitamin D index based solely on solar UVB doses related to latitude might be difficult.

Fortunately, there are other indices that can be used to estimate solar UV irradiance: skin cancer and melanoma mortality rates. Skin cancer deaths in Spain are largely due to squamous cell carcinoma (SCC), which is related to total lifetime solar UV irradiance.29 A recent analysis determined that diagnosis of nonmelanoma skin cancer (NMSC) is associated with reduced risk of several vitamin D-sensitive cancers as long as the confounding effects of smoking, a risk factor for skin cancer30, 31 and many other cancers,32, 33, 34 are considered.35 This study confirms results of a much earlier study that found an inverse relation between skin cancer and internal cancers.36 In that study, members of the U.S. Navy were compared with civilians; skin cancer rates for Navy personnel were 8 times that of the controls, but mortality rates from other cancers were 60% lower. Also, skin cancer was used as an index of solar UVB irradiance in a linkage study on risk of multiple sclerosis.37

Melanoma is another cancer linked to solar UV irradiance. However, it is linked to sunburning, limited solar UV irradiance,38 solar UVA irradiance,39, 40 low dietary vitamin D intake,5 lack of chronic solar UVB irradiation at higher latitudes6, 41 and a high-fat, low-fruit and vegetable diet.5 Melanoma has also been found associated with increased risk of lymphoma,42, 43 suggesting that UVA irradiance increased the risk of hematopoietic cancers through immunosuppression.

Also, lung cancer mortality rates can be used as an index of the health effects of smoking.4, 44, 45 Although lung cancer risk is also related to diet (intake of fat and animal products increases risk46) and vitamin D,15 one cannot separate these effects without knowing the geographic variation of long-term smoking prevalence.

Thus, the data in the Atlas of Cancer Mortality19 can be used without other input data in an ecologic study of the effects of solar UV irradiance and smoking on the risk of various forms of cancer. The results will also provide an interesting comparison of 3 indices of solar UV irradiance.

Material and methods

In this study, the primary goal was to examine the use of several indices of solar UV irradiance with respect to cancer in an ecologic study. The indices chosen are latitude and mortality rates for NMSC and melanoma. The latitude index is inversely correlated with solar UVB and vitamin D production potential for Spain. However, the effects of higher surface elevation in increasing UVB levels, and of clouds in reducing UVB levels, as well as residence and occupation, are not included in this index. NMSC mortality rates should be an index of higher population-average solar UVB irradiance over many years, either through living in regions with higher UVB doses or greater time spent out doors, such as in agricultural activities. Melanoma mortality rates could be an index of solar UV irradiance. However, there are several important risk-modifying factors for melanoma in addition to UV irradiance. Interestingly, melanoma mortality rates increase with increasing latitude for those living in their ancestral homelands, whereas rates decrease with increasing latitude for pale-skinned people who have migrated to countries such as Australia, New Zealand and the United States.47 This effect is probably because of the effect of skin pigmentation blocking the deep penetration of UVA as well as the variation of the UVB to UVA ratio with latitude,48, 49 although the effects of changing diet with latitude may also play a role [Grant, unpublished]. Thus, it is unclear how melanoma will be associated with other cancers.

Since smoking is a risk factor for NMSC and perhaps melanoma, an index of smoking should be used in the analysis. Lung cancer mortality rates have been demonstrated to be useful as an index of the health effects of smoking,4, 44 so it is included in this study.

The cancer mortality rate data for 52 provinces averaged for 1978–1992 were obtained from the Atlas of Cancer Mortality.19 Among the data tabulated are number of deaths in each province, cancer mortality rates adjusted to the European and world population age distributions and provincial rate ratios versus the national average. Data used in this work are the provincial rate ratios versus the average for each province. Populations for 2 provinces, Cueta and Melilla, which are near northern Africa, were low, often with single-digit deaths in the period. Also, 2 other provinces, Las Palmas and Santa Cruz, are in the Canary Islands, south of continental Spain. Thus, data for these 4 provinces are not included in the primary analysis.

The cross correlations among the 4 indices are given in Tables I and II. As can be seen, there is considerable cross-correlation between several of the factors. NMSC is inversely correlated with latitude, as would be expected [r = −0.50 (M), −0.33 (F)]. However, these correlation coefficients are not very high, indicating that factors other than latitude affect the risk of NMSC. Lung cancer is inversely correlated with latitude for males, suggesting that the effects of smoking also contribute to the latitudinal variations of cancer rates. Melanoma rates increase with increasing latitude at a marginally insignificant rate for females.

Table I. Linear Correlation Coefficient, r, and p Value for the Four Indices used in this Study with Age-Adjusted Mortality Rates Compared to the European Population Age Distribution for the 48 Continental Provinces for Males

Cancer	Mortality rate1	Latitude (r, p)	NMSC (r, p)	Melanoma (r, p)	Lung cancer (r, p)
* p from 0.01 to 0.05 ** p < 0.01; not stated p > 0.05. 1 Deaths/100,000/year.
Bladder	12.5	−0.37*	−0.03	0.40**	0.84**
Brain	5.44	0.30*	−0.31*	0.35*	0.38**
Buccal	7.61	−0.06	−0.18	0.26	0.77**
Colon	12.0	0.19	−0.30*	0.47**	0.52**
Esophageal	7.95	0.27	−0.30*	0.07	0.56**
Gallbladder	1.80	0.16	−0.31*	0.36*	0.02
Gastric	25.0	0.41**	0.01	−0.39**	−0.40**
Hodgkin's lymphoma	1.13	0.007	−0.12	0.16	0.35*
Laryngeal	10.5	−0.34*	0.13	0.15	0.81**
Leukemia	6.71	−0.14	−0.04,	0.27	0.51**
Lung	60.2	−0.36*	0.02	0.33*
Melanoma	1.01	−0.08	−0.03		0.33*
Multiple myeloma	2.18	0.17	−0.24	0.57**	0.31*
NHL	3.58	0.24	−0.32*	0.49**	0.51**
NMSC	1.63	−0.50**		−0.03	0.02
Pancreatic	7.51	0.55**	−0.35*	0.24	0.40**
Peritoneal	0.96	−0.02	−0.23	0.27	0.32*
Pleural	0.55	0.13	−0.40**	0.43**	0.37*
Prostate	21.6	0.06	−0.21	0.52**	0.49**
Rectal	6.69	0.61**	−0.46**	0.31*	−0.03
Renal	3.54	0.22	−0.18	0.34*	0.22
Thyroid	0.37	0.32*	−0.35*	0.09	0.24

Table II. Linear Correlation Coefficient, r, and p Value for the Four Indices used in this Study with Age-Adjusted Mortality Rates Compared to the European Population Age Distribution for the 48 Continental Provinces for Females

Cancer	Mortality rate1	Latitude (r, p)	NMSC (r, p)	Melanoma (r, p)	Lung cancer (r, p)
* p from 0.01 to 0.05 ** p < 0.01; not stated p > 0.05. 1 Deaths/100,000/year.
Bladder	1.89	−0.38*	−0.29*	0.18	0.34*
Brain	3.37	0.58**	−0.46**	0.29*	0.05
Breast	21.6	0.15	−0.38**	0.30*	0.23
Buccal	1.04	0.006	−0.11	0.21	0.14
Colon	8.99	−0.006	−0.31*	0.43**	0.38**
Esophageal	1.10	−0.25	0.09	−0.07	0.37*
Gallbladder	2.95	−0.18	−0.20	0.12	−0.15
Gastric	12.3	0.46**	0.23	−0.25	−0.37*
Hodgkin's lymphoma	0.58	0.02	−0.47**	0.17	0.41**
Leukemia	4.26	−0.28	0.11	0.30*	0.40**
Lung	5.44	−0.05	−0.31*	0.36*
Melanoma	0.70	0.28	−0.43**		0.36*
Multiple myeloma	1.57	0.10	−0.45**	0.58**	0.26
NHL	2.15	0.29*	−0.63**	0.61**	0.40**
NMSC	0.79	−0.33*		−0.43**	−0.31*
Ovarian	4.30	0.21	−0.51**	0.64**	0.47**
Pancreatic	4.42	0.40**	−0.35*	0.45**	0.28
Peritoneal	1.02	−0.26	−0.23	0.22	0.25
Pleural	0.26	0.29*	−0.41**	0.002	0.12
Rectal	4.14	0.33*	−0.35*	0.26	0.04
Renal	1.45	0.13	−0.11	0.27	0.07
Thyroid	0.59	0.48**	−0.31*	0.00	0.14
Uterine corpus	8.62	−0.40**	−0.29*	0.29*	0.55**

The data for each province were compared in linear regression analyses for each cancer and sex by using SPSS 13.0 (SPSS, Chicago, IL). No adjustment was made for the number of cancer deaths or the populations of each province other than to omit the non-continental provinces as mentioned.

Results

Results were omitted from the tables for cancers for which no significant results for any UV indices were found. The linear regression results for the continental provinces are presented in Tables I and II. The results using the 4 noncontinental provinces are considered less reliable than those using just the continental provinces, in part since the populations in those provinces are low. However, the results including all 52 provinces did not differ significantly from those using the 48 continental provinces. Thus, it appears justified to concentrate on the results for the 48 continental provinces.

The statistically significant results are summarized here for the regression results in Tables I and II. The 9 cancers with mortality rates (M, males; F, females) significantly correlated with latitude are brain, gastric, melanoma (F), NMSC, non-Hodgkin's lymphoma (NHL; F), pancreatic, pleural (F), rectal and thyroid cancer. Inverse correlations with latitude were found for laryngeal (M), lung (M) and uterine corpus cancer. The 17 cancers inversely correlated with NMSC [9 (M), 15 (F)] are bladder (F), brain, breast (F), colon, esophageal (M), gallbladder (M), Hodgkin's lymphoma (F), lung (F), melanoma (F) multiple myeloma (F), NHL, ovarian, pancreatic (M), pleural, rectal, thyroid and uterine corpus cancer. The 16 correlated with melanoma are bladder (M), brain, breast (F), colon, gallbladder (M), leukemia (F), lung, multiple myeloma, NHL, ovarian, pancreatic (F), pleural (M), prostate, rectal (M), renal (M) and uterine corpus cancer. Gastric cancer was inversely correlated with melanoma.

Lung cancer mortality rates for males are 11 times higher than for females. There are 15 cancer sites for which lung cancer is significantly associated with other cancers for males, but only 10 for females, for a total of 17 sites with significant correlations. Two cancers, gastric and NMSC (F), were inversely correlated with lung cancer.

Discussion

The inverse correlations with NMSC as a proxy for high solar UVB irradiance as a proxy for low vitamin D serum levels agree with other findings identifying vitamin D-sensitive cancers.2, 3, 4 One other study reported a correlation with latitude for multiple myeloma.50 However, several cancers identified as UVB/vitamin D sensitive in other studies were not confirmed in this study such as bladder, laryngeal and renal cancer and leukemia. In such cases, the risk reduction by vitamin D is thought to be a less significant factor than other important risk factors such as smoking and diet. The findings for brain, pleural and thyroid cancer may be new. Another study reported that vitamin D supplement use was protective against thyroid cancer.51 More studies to confirm these findings are indicated.

In the analyses, latitude yielded stronger correlations with cancer mortality rates than did NMSC mortality rates for gastric, pancreatic, rectal (M) and thyroid (M) cancer, whereas NMSC yielded stronger associations for breast, colon, multiple myeloma, NHL, ovarian, and pleural cancer. Strong inverse associations with latitude were found for bladder, lung (M), NMSC and uterine corpus cancer. Thus, latitude is an index that includes the effects of both solar UVB and smoking, so cannot be used reliably as an index of either factor.

Gastric cancer appears to have other important risk factors that were not modeled, such as diet. Since many factors were not included, single-factor analyses are prone to error. However, this study shows that NMSC mortality rates are an important measure of solar UVB irradiance at the population level and that use of this index yields results in an ecologic study that generally confirms findings of other ecologic studies that used geographic location with respect to solar UVB doses.

Nonetheless, the results do indicate that the results for the effect of indices related to vitamin D production are similar to those reported in other countries. That NMSC more often yields better correlations than latitude for females than males could indicate that females spend less time out doors than do males, so that when those in various provinces do spend more time out doors, there is enhanced risk of NMSC but lower risk of internal cancers. However, when latitude has a higher correlation than does NMSC for males, smoking often has the highest correlation with cancer mortality rates, suggesting that confounding affects the results. This finding was also seen in the analysis of papers reporting development of a second cancer after diagnosis of NMSC.35

All the cancers that are correlated with melanoma have been identified as vitamin D sensitive except brain and pleural cancer and multiple myeloma, although an increase with latitude was noted for multiple myeloma in an ecologic study using European countries.50 There is a large body of literature presenting observational studies reporting associations between melanoma and other cancers at the individual level.52, 53, 54, 55, 56, 57, 58, 59 Thus, melanoma probably shares many risk factors with other cancers, so it cannot be concluded from the results in this work that vitamin D accounts for the association. It could also be the case that average skin pigmentation decreases with increasing latitude.

However, the hematopoietic cancers (leukemia, multiple myeloma and NHL) are associated with immunosuppression from UVA.43, 60 Thus, for these cancers, the associations found may be due to a combination of vitamin D production and immunosuppression.

At the individual level, diagnosis of SCC is associated with increased risk of internal cancers.61, 62 This finding may be associated with risk factors other than solar UVB irradiance, such as smoking and dietary factors, as discussed in other article.35 Thus, use of NMSC mortality rates as an index of solar UVB irradiance appears to work better at the population level, rather than at the individual level.

That NMSC mortality rates are generally inversely correlated with the cancers in this study, whereas those for melanoma are generally positively correlated, provides more evidence that the etiologies of SCC and melanoma are different. SCC is linked to cumulative solar UVB irradiance29 and smoking,30, 31 whereas melanoma is more strongly correlated with UVA irradiance,39, 40 inversely correlated with chronic solar UV irradiance for those whose skin is appropriate for the ambient solar UV doses,6, 41 and dietary vitamin D,5 and apparently possibly correlated with smoking.35, 63 Thus, guidelines for reducing the risk of melanoma and SCC should probably be different.

Lung cancer was inversely correlated with NMSC for females but not correlated with NMSC for males, suggesting that smoking does not affect the risk of NMSC in Spain. The inverse correlation for females may be coincidental or it may indicate a protective effect of solar UVB irradiance. Lung cancer is correlated with melanoma, suggesting that either smoking and/or the diet associated with smoking could play a role in the etiology of melanoma in Spain, although confounding by other factors cannot be ruled out.

Although the NMSC mortality rates generally have similar correlations with other cancers, as does latitude, the correlation coefficients are not particularly high. One reason was that many factors that affect risk of cancer were not included in the analysis. A second reason is that the variation of solar UVB doses are in the range of 10–30%.

Of those cancers for which the association with lung cancer was significant, many have been linked to smoking.32, 33, 34 Although not generally included in such listings, there is also support for smoking as a risk factor for Hodgkin's lymphoma64, 65 and ovarian cancer.66

The results for uterine corpus cancer are puzzling: latitude is inversely correlated with mortality rates, although uterine corpus cancer is inversely correlated with NMSC and positively so with melanoma. Evidently, unmodeled factors are important.

Smoking is not generally linked to prostate cancer other than reducing risk by reducing life expectancy. Smoking has long been associated inversely with uterine corpus cancer.67 However, even if smoking is not a direct risk factor for uterine corpus cancer, the finding could be related to smoking through diet or the effect on vitamin D and calcium. Smokers in Spain and elsewhere generally have less healthy diets than those of nonsmokers.68 Smoking is associated with reduced vitamin D serum levels69, 70 and calcium availability.71

These results provide more support for the UVB/vitamin D/cancer hypothesis and are generally in good agreement with findings reported elsewhere, thus strongly indicating that cancer mortality rates in Spain could be reduced considerably if people obtained more vitamin D from both solar UVB and oral intake D and smoked less.

This study is ecologic, with 3 indices associated with solar UVB irradiance and 1 for smoking, and has some limitations. There are many more cancer risk-modifying factors that were not included in the analysis that could affect the findings, such as diet, alcohol consumption, socioeconomic status and skin pigmentation. However, on the basis of ecologic studies of cancer mortality rates in the United States,2, 4, 45 these results should be reasonably correct regarding the contributions of solar UVB and smoking. Nonetheless, extending this study by incorporating such factors in the analysis would be useful.

Summary and Conclusion

Although no one should be encouraged to risk developing skin cancer, the results presented here indicate that the risks associated with solar UVB irradiance are overshadowed by the health benefits of reduced risk of internal cancers. Also, there are benefits for many other conditions and diseases, including autoimmune diseases, cardiovascular diseases, bones and muscles etc.72, 73, 74, 75, 76 A recent analysis for the United States estimated that the economic burden of insufficient UVB and vitamin D was nearly an order of magnitude greater than the economic burden of excess solar UV.77

Thus, Spain and other European countries should further investigate the benefits of solar UVB and vitamin D and modify their guidelines for food fortification,78 supplements and solar UV irradiance as deemed appropriate. Such steps are under way in Australia and New Zealand79 and North America,80 although the recommended daily vitamin D levels, 400 IU, are too low for optimal health and cancer prevention,3, 11, 81, 82 having been based on older, outdated recommendations for bone health.83 Increasing vitamin D production and oral intake for all to 1000–1500 I.U./day in Spain would likely reduce cancer rates by 15–30%.3, 11, 84 Additional support for the melanoma results presented in this work is given in a case control study of melanoma risk with respect to sun exposure in Spain.85