We are happy to reply to the points raised by William B. Grant (1), who states that UVB exposure (rather than circadian disruption) may account for our findings of increased cancer rates associated with western position in U.S. time zones based on an ecologic analysis we conducted using SEER data (2).
First, some corrections to Dr. Grant's text (1). Although we included latitude in our models, our findings implicate westerly position in the time zone or longitude, not latitude, as a risk factor, consistent with circadian disruption. Latitude alters day length, angle of the sun, temperature, and other factors and therefore all our analyses included it as a potentially confounding variable. The shift in longitude for each time zone is 15 degrees (not 20). The adjusted rate ratio elevations for all cancers we reported are for 20 minutes (=5 degrees longitude) and were close to 4% in females and 3% in men.
Second, we disagree with Dr. Grant's assertion that the findings are random with respect to individual cancers. For example, risks associated with female breast (3), prostate (4), uterine corpus (5), and chronic lymphocytic leukemia (CLL) (6) are consistent with the previous literature that reported elevated risk in shift workers.
Third, given our broad findings involving multiple cancers, we chose to include only a limited number of important covariates in our analyses. Future studies focusing on individual cancers should emphasize specific confounders as appropriate, although information on many potential confounders is not routinely available at the county level. Conducting analytic epidemiology studies in populations where individual data can be incorporated will be an important next step.
Finally, we conducted the analyses suggested by Dr. Grant. We excluded California and Washington counties and limited the analysis to non-Hispanic whites. The findings are similar to Table 1 from our original work (1) with some attenuation due to the reduction in sample size. Next, using the full data, we controlled for UVB using data based on county-level summertime ambient UVB estimates. These results again were similar to our original findings (1). While some of the results are attenuated consistent with a degree of collinearity between sunlight (which drives the daily reset of the biological clock in humans) and UVB exposure, we note that Bonferroni thresholds for statistical significance are achieved for all cancers (both genders), prostate, uterine corpus, and CLL in men among others (Table 1). While there are ecologic and observational epidemiology data associating UV radiation exposure with both increased (melanoma) and decreased (non–Hodgkin lymphoma, colon, others) cancer incidence (7), additional study including mediation analyses are needed to investigate alternative potential mechanisms such as vitamin D, activity/inactivity, sleep alterations, obesity and insulin resistance.
. | Male . | Female . | ||||
---|---|---|---|---|---|---|
Cancerd . | Cases . | RR (95% CI)g . | P . | Cases . | RR (95% CI)g . | P . |
All cancers | 2,095,394 | 1.022 (1.010–1.034) | 2.6 × 10−4 | 1,972,514 | 1.034 (1.021–1.046) | 6.4 × 10−8 |
Oral cavity and pharynxe | 54,889 | 1.018 (0.979–1.059) | 0.36 | 23,199 | 1.064 (1.008–1.123) | 0.02 |
Esophagus | 30,956 | 1.003 (0.959–1.050) | 0.88 | 8,751 | 1.162 (1.072–1.259) | 2.8 × 10−4 |
Adenocarcinoma | 21,355 | 1.008 (0.954–1.066) | 0.77 | 3,448 | 1.078 (0.965–1.204) | 0.19 |
Squamous cell | 6,570 | 0.993 (0.912–1.081) | 0.88 | 4,150 | 1.029 (0.916–1.156) | 0.63 |
Stomach | 35,990 | 1.087 (1.033–1.143) | 1.3 × 10−3 | 21,924 | 1.048 (0.981–1.118) | 0.16 |
Colon and rectum | 205,056 | 1.021 (1.000–1.042) | 0.05 | 193,349 | 1.044 (1.023–1.065) | 3.6 × 10−5 |
Liver and intrahepatic bile duct | 39,276 | 1.092 (1.028–1.159) | 4.3 × 10−3 | 15,796 | 1.093 (1.021–1.171) | 0.01 |
Gallbladder | 2,834 | 0.887 (0.771–1.021) | 0.10 | 6,683 | 1.037 (0.949–1.134) | 0.42 |
Pancreas | 51,150 | 1.035 (1.002–1.070) | 0.04 | 50,553 | 1.045 (1.010–1.082) | 0.01 |
Larynx | 24,578 | 0.985 (0.939–1.034) | 0.55 | 6,158 | 1.058 (0.977–1.146) | 0.16 |
Lung and bronchus | 288,452 | 0.996 (0.971–1.023) | 0.79 | 256,618 | 1.041 (1.012–1.071) | 6.0 × 10−3 |
Adenocarcinoma | 89,166 | 1.005 (0.965–1.047) | 0.79 | 94,771 | 1.052 (1.012–1.094) | 0.01 |
Squamous cell | 65,665 | 1.011 (0.971–1.052) | 0.60 | 38,324 | 1.048 (0.991–1.108) | 0.10 |
Small cell | 35,852 | 0.987 (0.942–1.034) | 0.58 | 35,834 | 1.040 (0.988–1.094) | 0.13 |
Melanoma of the skin | 119,304 | 1.040 (0.989–1.095) | 0.13 | 85,934 | 1.004 (0.948–1.063) | 0.90 |
Breast | 4,463 | 1.073 (0.968–1.189) | 0.18 | 593,753 | 1.029 (1.011–1.047) | 1.8 × 10−3 |
HR positive | 3,713 | 1.022 (0.915–1.141) | 0.70 | 435,649 | 1.037 (1.009–1.066) | 0.01 |
HR negative | 141 | 0.699 (0.185–2.644) | 0.60 | 92,225 | 1.030 (0.999–1.061) | 0.06 |
Cervix uteri | 0 | — | — | 34,283 | 1.052 (1.004–1.102) | 0.03 |
Corpus and uterus NOSf | 0 | — | — | 113,344 | 1.086 (1.053–1.121) | 2.7 × 10−7 |
Ovary | 0 | — | — | 62,958 | 1.006 (0.972–1.041) | 0.74 |
Prostate | 578,119 | 1.042 (1.018–1.066) | 6.0 × 10−4 | 0 | — | — |
Urinary bladder | 147,266 | 1.005 (0.979–1.032) | 0.70 | 47,073 | 1.016 (0.973–1.060) | 0.48 |
Kidney and renal pelvis | 80,453 | 1.025 (0.994–1.058) | 0.12 | 47,964 | 1.018 (0.980–1.057) | 0.36 |
Brain | 32,043 | 1.009 (0.968–1.052) | 0.66 | 24,808 | 0.996 (0.949–1.045) | 0.87 |
Thyroid | 24,274 | 1.064 (0.996–1.137) | 0.07 | 73,382 | 1.085 (1.028–1.144) | 3.1 × 10−3 |
Non-Hodgkin lymphoma | 95,143 | 1.033 (1.004–1.064) | 0.03 | 80,861 | 1.030 (1.001–1.060) | 0.05 |
Myeloma | 26,959 | 0.968 (0.923–1.015) | 0.18 | 20,839 | 1.060 (1.008–1.115) | 0.02 |
Chronic lymphocytic leukemia | 25,251 | 1.112 (1.054–1.172) | 1.0 × 10−4 | 17,072 | 1.087 (1.022–1.156) | 7.9 × 10−3 |
Acute myeloid leukemia | 17,990 | 1.033 (0.977–1.093) | 0.25 | 14,815 | 1.055 (0.995–1.118) | 0.07 |
. | Male . | Female . | ||||
---|---|---|---|---|---|---|
Cancerd . | Cases . | RR (95% CI)g . | P . | Cases . | RR (95% CI)g . | P . |
All cancers | 2,095,394 | 1.022 (1.010–1.034) | 2.6 × 10−4 | 1,972,514 | 1.034 (1.021–1.046) | 6.4 × 10−8 |
Oral cavity and pharynxe | 54,889 | 1.018 (0.979–1.059) | 0.36 | 23,199 | 1.064 (1.008–1.123) | 0.02 |
Esophagus | 30,956 | 1.003 (0.959–1.050) | 0.88 | 8,751 | 1.162 (1.072–1.259) | 2.8 × 10−4 |
Adenocarcinoma | 21,355 | 1.008 (0.954–1.066) | 0.77 | 3,448 | 1.078 (0.965–1.204) | 0.19 |
Squamous cell | 6,570 | 0.993 (0.912–1.081) | 0.88 | 4,150 | 1.029 (0.916–1.156) | 0.63 |
Stomach | 35,990 | 1.087 (1.033–1.143) | 1.3 × 10−3 | 21,924 | 1.048 (0.981–1.118) | 0.16 |
Colon and rectum | 205,056 | 1.021 (1.000–1.042) | 0.05 | 193,349 | 1.044 (1.023–1.065) | 3.6 × 10−5 |
Liver and intrahepatic bile duct | 39,276 | 1.092 (1.028–1.159) | 4.3 × 10−3 | 15,796 | 1.093 (1.021–1.171) | 0.01 |
Gallbladder | 2,834 | 0.887 (0.771–1.021) | 0.10 | 6,683 | 1.037 (0.949–1.134) | 0.42 |
Pancreas | 51,150 | 1.035 (1.002–1.070) | 0.04 | 50,553 | 1.045 (1.010–1.082) | 0.01 |
Larynx | 24,578 | 0.985 (0.939–1.034) | 0.55 | 6,158 | 1.058 (0.977–1.146) | 0.16 |
Lung and bronchus | 288,452 | 0.996 (0.971–1.023) | 0.79 | 256,618 | 1.041 (1.012–1.071) | 6.0 × 10−3 |
Adenocarcinoma | 89,166 | 1.005 (0.965–1.047) | 0.79 | 94,771 | 1.052 (1.012–1.094) | 0.01 |
Squamous cell | 65,665 | 1.011 (0.971–1.052) | 0.60 | 38,324 | 1.048 (0.991–1.108) | 0.10 |
Small cell | 35,852 | 0.987 (0.942–1.034) | 0.58 | 35,834 | 1.040 (0.988–1.094) | 0.13 |
Melanoma of the skin | 119,304 | 1.040 (0.989–1.095) | 0.13 | 85,934 | 1.004 (0.948–1.063) | 0.90 |
Breast | 4,463 | 1.073 (0.968–1.189) | 0.18 | 593,753 | 1.029 (1.011–1.047) | 1.8 × 10−3 |
HR positive | 3,713 | 1.022 (0.915–1.141) | 0.70 | 435,649 | 1.037 (1.009–1.066) | 0.01 |
HR negative | 141 | 0.699 (0.185–2.644) | 0.60 | 92,225 | 1.030 (0.999–1.061) | 0.06 |
Cervix uteri | 0 | — | — | 34,283 | 1.052 (1.004–1.102) | 0.03 |
Corpus and uterus NOSf | 0 | — | — | 113,344 | 1.086 (1.053–1.121) | 2.7 × 10−7 |
Ovary | 0 | — | — | 62,958 | 1.006 (0.972–1.041) | 0.74 |
Prostate | 578,119 | 1.042 (1.018–1.066) | 6.0 × 10−4 | 0 | — | — |
Urinary bladder | 147,266 | 1.005 (0.979–1.032) | 0.70 | 47,073 | 1.016 (0.973–1.060) | 0.48 |
Kidney and renal pelvis | 80,453 | 1.025 (0.994–1.058) | 0.12 | 47,964 | 1.018 (0.980–1.057) | 0.36 |
Brain | 32,043 | 1.009 (0.968–1.052) | 0.66 | 24,808 | 0.996 (0.949–1.045) | 0.87 |
Thyroid | 24,274 | 1.064 (0.996–1.137) | 0.07 | 73,382 | 1.085 (1.028–1.144) | 3.1 × 10−3 |
Non-Hodgkin lymphoma | 95,143 | 1.033 (1.004–1.064) | 0.03 | 80,861 | 1.030 (1.001–1.060) | 0.05 |
Myeloma | 26,959 | 0.968 (0.923–1.015) | 0.18 | 20,839 | 1.060 (1.008–1.115) | 0.02 |
Chronic lymphocytic leukemia | 25,251 | 1.112 (1.054–1.172) | 1.0 × 10−4 | 17,072 | 1.087 (1.022–1.156) | 7.9 × 10−3 |
Acute myeloid leukemia | 17,990 | 1.033 (0.977–1.093) | 0.25 | 14,815 | 1.055 (0.995–1.118) | 0.07 |
NOTE: P values less than 1.8 × 10−3 (0.05/28) for men and less than 1.7 × 10−3 (0.05/30) for women are in bold and indicate statistical significance using Bonferroni correction.
Abbreviations: CI, confidence interval.
aThe position in the time zone (PTZ) is calculated as the distance (in degrees longitude) between the population centroid of each county and the middle meridian of the located time zone (EST:75, CST:90, MST:105, and WST:120).
bThe ambient UVB exposure is derived using the Total Ozone Mapping Spectrometer (TOMS) database maintained by the National Aeronautics and Space Administration (NASA). Cloud-adjusted daily ambient ultraviolet irradiance at 305 nm is provided on a 1 latitude × 1 longitude grid.
cA weighted (by population size) logarithmic linear regression was used between PTZ and age-adjusted (2000 US standard population) county-level cancer incidence rate. Model adjusted for latitude, poverty percentage, smoking, state, and summertime UVB.
dOrdered by organ systems, based on SEER Statistics Review 1975–2013.
eExcluding lip and salivary glands.
fNot otherwise specified.
gRate ratios are per five degrees of longitude difference, equivalent to 20 minutes.
In conclusion, based on the previous findings (1) and these additional analyses, circadian disruption remains the most plausible explanation for the elevated relative risk of diverse cancers in the western regions of time zones in the United States. Given the broad impact of circadian biology on human health (8) and accelerating genetic (9, 10) understanding, further population-based studies to investigate its impact are a priority.
See the original Letter to the Editor, p. 1110
Disclosure of Potential Conflicts of Interest
E.B. Klerman is a consultant/advisory board member for Pfizer Pharmaceuticals and has provided expert testimony for Sleep Research Society. No potential conflicts of interest were disclosed by the other authors.