Seasonal variation in the patient diagnosis of ... - Wiley Online Library

AUTOIMMUNE, CHOLESTATIC AND BILIARY DISEASE

Seasonal Variation in the Patient Diagnosis of Primary Biliary Cirrhosis: Further Evidence for an Environmental Component to Etiology Richard J.Q. McNally,1 Peter W. James,1 Samantha Ducker,2 and Oliver F.W. James2 The etiology of primary biliary cirrhosis (PBC) is far from clear. Both genetic and environmental factors are likely to be involved. We have previously reported evidence of space-time clustering, suggesting that a transient environmental agent may be involved in etiology. To further examine whether a seasonally varying environmental agent may contribute to the etiology of PBC, we have analyzed seasonal variation with respect to month of diagnosis using population-based data from northeast England over a defined period (1987-2003). Date of diagnosis was defined as the earliest date at which the patient was found to have fulfilled any two of three diagnostic criteria (i.e., antimitochondrial antibody–positive titer 1 in 40, cholestatic liver blood tests, diagnostic or compatible liver histology). Monthly expected (E) numbers of cases were calculated under an assumption of a uniform distribution throughout the year. Observed counts (O) were compared with the expected numbers. The chi-squared heterogeneity test was used to test for overall nonuniform variation and also for individual months. Poisson regression analysis was used to fit a sinusoidal (i.e., harmonic) model to the data, using month of diagnosis as a covariate in the model. There was a marked peak for diagnoses in the month of June (O 5 115, E 5 84.7, O/E 5 1.36; P 5 0.001). Furthermore, there was evidence of a sinusoidal pattern with a June peak (P 5 0.012). Conclusion: These highly novel results provide further evidence for the involvement of a seasonally varying environmental agent in the etiology of PBC. (HEPATOLOGY 2011;54:2099-2103)

T

he etiology of primary biliary cirrhosis (PBC) is not clear.1 Both genetic2-4 and environmental factors are likely to be involved. We have previously reported evidence of space-time clustering among cases of PBC in a defined geographical population of northeast England.5 This finding suggested that one or more transient environmental agents may play a role in etiology. Putative agents, suggested by other studies, include infections, such as Escherichia coli, mycobacteria, and a retrovirus.6-9 An earlier small study from northeast England of 117 cases of PBC diagnosed during 1966Abbreviations: AMA, antimitochondrial antibody; E, expected number of cases; ICD, International Classification of Diseases; O, observed number of cases; ONS, Office for National Statistics; PBC, primary biliary cirrhosis. From the 1Institute of Health and Society and 2Institute of Cellular Medicine, Newcastle University, Newcastle upon Tyne, United Kingdom. Received March 2, 2011; accepted July 22, 2011. This study was supported by the Medical Research Council (UK). Address reprint requests to: Richard J.Q. McNally, Ph.D., Institute of Health and Society, Newcastle University, Sir James Spence Institute, Royal Victoria Infirmary, Queen Victoria Road, Newcastle upon Tyne NE1 4LP, United Kingdom. E-mail: [email protected]; fax: þ44-191-282-4724. C 2011 by the American Association for the Study of Liver Diseases. Copyright V View this article online at wileyonlinelibrary.com. DOI 10.1002/hep.24597 Potential conflict of interest: Nothing to report.

1979 had shown evidence of seasonality in symptom development, particularly in the spring and early summer,10 although this finding has never been confirmed. If seasonally varying transient environmental agents contribute to the etiology of a disease, then the distribution of cases may exhibit seasonal patterning. However, such seasonality would only happen under very specific conditions. In the case of PBC, this would imply the following: (1) the agent would have a seasonal pattern of occurrence; (2) the latent period from exposure to diagnosis would be relatively constant; and (3) because PBC is a relatively uncommon disease, the onset of PBC would result as a rare consequence of exposure to the transient environmental agent. Examples of agents that may exhibit a seasonal pattern include infections, air pollution, and dietary factors. The aim of the present study was to investigate seasonal variation in the incidence of PBC by month of diagnosis among cases diagnosed during 1987-2003 in a well-defined geographical area of northeast England.

Patients and Methods Cases: Case Definition. For this study, we included both cases defined as ‘‘definite PBC’’ and ‘‘probable 2099

2100

MCNALLY ET AL.

PBC’’ in our original case-finding study.11 Definite PBC is all three of the following: antimitochondrial antibody (AMA) positive titer 1 in 40, cholestatic liver blood tests, and diagnostic or compatible liver histology. Probable PBC is any two of the above three criteria (usually AMA-positive 1 in 40 and cholestatic liver blood tests in the absence of liver biopsy). These criteria are now widely accepted.1 For this reason, we refer to all cases with the above-described criteria—either ‘‘definite’’ or ‘‘probable’’ as ‘‘cases.’’ We defined ‘‘symptomatic’’ patients as those with pruritus, persistent fatigue, or signs and symptoms of cirrhosis. Patients with none of these were regarded as ‘‘asymptomatic’’ of liver disease at diagnosis. Time Frame and Study Area. The study included all cases incident between January 1, 1987 and December 31, 2003 and who were resident in an area of northeast England (i.e., Northumberland, Sunderland, North Durham, South Durham, Newcastle upon Tyne, North Tyneside, South Tyneside, and Gateshead), defined by postal (ZIP) code. The total population of the area at the 2001 census was less than 2.05 million.12 Case Finding. The methods for case finding have been described previously.13 Briefly, they were as follows: 1. Requests were made to all gastroenterologists and hepatologists in the region to identify all cases of PBC under their care. 2. Hospital admission data on Regional Information Systems for all 13 hospitals in the region using International Classification of Diseases (ICD)-9 code 571.6 (to April 1994) and ICD-10 code K74-3 thereafter were examined. 3. All hospital immunology laboratory data for patients with positive AMA 1 in 40 by indirect immunofluorescence were examined (over 500,000 laboratory records examined). 4. All listings from the Office for National Statistics (ONS) of deaths within the region and study period in which PBC, ICD-9 code 571.6, or (subsequently) ICD-10 code K74.3 appeared anywhere on the death certificate were examined. Case selection was approved by the local ethical committees. After initial identification, hospital records of all cases were reviewed. Date of Diagnosis. Date of diagnosis was defined as the earliest date at which the patient was found (by examination of clinical case records—hospital or primary care) to have fulfilled any two of the three diagnostic criteria. This was to avoid the need for different criteria for date of diagnosis in the asymptomatic versus the symptomatic group of patients. It is emphasised, therefore, that date of diagnosis was not the date

HEPATOLOGY, December 2011

at which a diagnosis of PBC was first made and entered in an individual’s clinical case records by the attending doctors.11 Rather, date of diagnosis was determined after examination by the investigators of clinical records and depended upon the date at which the above diagnostic criteria were first fulfilled. Prior Hypothesis. The following etiological hypothesis was tested: A primary factor influencing temporal heterogeneity of PBC is related to exposure to a seasonally varying environmental agent occurring close to diagnosis or at similar times before diagnosis. Statistical Methods. Monthly expected (E) numbers of cases were calculated under an assumption of a uniform distribution throughout the year. Observed counts (O) were compared with the expected numbers. A chi-squared test for heterogeneity was used to test for an overall seasonal effect in incidence. The test shows the presence of any departure from a uniform distribution throughout the year. Individual chisquared tests for each month were used to test for the presence of specific excesses. Poisson analysis was used to determine the pattern of seasonality. A sinusoidal (i.e., harmonic) model was fitted to the data, using month of diagnosis as a covariate. The Poisson model used was of the following form: p p EðOi Þ ln ¼ A sin i þ B cos i ; i ¼ 1; 2…12 Di 6 6 where Oi are the observed monthly counts, Di are the expected monthly counts assuming no seasonality, E(Oi) are the expected monthly counts assuming sinusoidal seasonal variation, and A and B are constants estimated from using the data.14 The month at which the maximum incidence occurred was estimated using a single annual peak within the period of 1 year. The significance of the sinusoidal model assumption was evaluated by using the Pearson chi-squared goodness-of-fit test.15 Statistical significance was taken to be P < 0.05, and marginal significance was taken as 0.05 P < 0.10. All statistical analyses were performed using STATA version 10 (StatCorp LP, College Station, TX).

Results The study analyzed date of diagnosis of 1,030 cases of PBC diagnosed during January 1, 1987 to December 31, 2003 within the specified region and whose details were extracted from the population-based register. There were 931 females and 99 males included in this analysis. Among the first 770 cases, in whom this was available, the diagnosis of PBC (and hence its

HEPATOLOGY, Vol. 54, No. 6, 2011

MCNALLY ET AL.

2101

Table 1. Results of Chi-Squared Tests of Seasonal Variation Performed on Monthly Frequencies for Cases of PBC Diagnosed During 1987-2003 Month

Observed (O)

Expected (E)

O/E

Chi-Squared Statistic

P Value

January February March April May June July August September October November December Total

71 74 85 83 93 115 94 77 88 78 92 80 1,030

87.5 79.0 87.5 84.7 87.5 84.7 87.5 87.5 84.7 87.5 84.7 87.5 1,030

0.81 0.94 0.97 0.98 1.06 1.36 1.07 0.88 1.04 0.89 1.09 0.91 1.00

3.10 0.32 0.07 0.03 0.35 10.88 0.49 1.26 0.13 1.03 0.64 0.64 18.93

0.078 0.573 0.791 0.857 0.555 0.001 0.486 0.263 0.716 0.311 0.425 0.424 0.062

date) was made retrospectively by the investigators in 34.3% (264) of patients. Twelve cases were uniquely detected by death certificate, and these were not included in the 1,030 cases that were analyzed. The overall chi-squared heterogeneity test showed marginally statistically significant evidence for departure from the uniform distribution (P ¼ 0.062). More detailed analyses of individual months showed that there was a marked, highly statistically significant peak for diagnoses in the month of June (O ¼ 115, E ¼ 84.7, O/E ¼ 1.36; P ¼ 0.001). Furthermore, Poisson modeling showed that there was evidence of sinusoidal patterning with a June peak (P ¼ 0.012), with goodness of fit (P ¼ 0.302) and estimated amplitude of 0.111 (standard error ¼ 0.044) (Table 1; Fig. 1). The numbers of patients attending liver clinics and inpatient admissions by month during 2001-2003 are given in Table 2. Although there were lower attendances during August and December because of staff vacations, there was no evidence of any seasonal patterning (P ¼ 0.712).

The numbers of patients by month surviving less than 5 years (i.e., late diagnoses) and numbers of patients surviving more than 10 years (i.e., early diagnoses) are given in Table 3. There was evidence of seasonality among the early diagnoses (chi-square test for heterogeneity: P ¼ 0.035; test for sinusoidal variation: P ¼ 0.011). However, there was no evidence of seasonality among the late diagnoses (chi-square test for heterogeneity: P ¼ 0.781; test for sinusoidal variation: P ¼ 0.780). The numbers of symptomatic and asymptomatic patients by month of diagnosis are given in Table 4. There was evidence of significant sinusoidal variation among the symptomatic group (P ¼ 0.013). There was evidence of significant heterogeneity among the asymptomatic group (P ¼ 0.004) (Fig. 2).

Discussion This study has found highly novel evidence of seasonal variation among cases of PBC. Rigorous statistical methods have been used to analyze high-quality, population-based data from a well-defined geographical Table 2. Liver Clinic Attendances and Inpatient Admissions by Month During 2001-2003*

Fig. 1. Plot of raw data and sinusoidal model of seasonal variation in monthly frequencies for cases of PBC diagnosed during 19872003.

Month

Liver Clinic Attendances and Inpatient Admissions


368 381 374 357 383 377 356 334 382 388 378 328 4,406

*Attendance lower during August and December because of staff vacations.

2102

MCNALLY ET AL.

HEPATOLOGY, December 2011

Table 3. Numbers of Patients by Month Surviving Less Than 5 Years (Late Diagnoses) and Number of Cases Surviving More Than 10 Years (Early Diagnoses)*

Month

Numbers of Patients Surviving Less Than 5 Years (Late Diagnoses)

Numbers of Patients Surviving Greater Than 10 Years (Early Diagnoses)


8 10 4 9 11 9 9 8 5 10 9 5 97

49 54 57 64 66 87 70 60 65 50 72 59 753

*Ddata are incomplete for 10-year follow-up in patients included after 2000.

region. The study area has very low inward or outward migration rates.12,16,17 Thus, the findings cannot be explained by seasonal migration. We have also reviewed our case-finding methodology very carefully and can find no reason within this for a seasonal variation in notification of new cases over this 17-year period. Case finding was carried out in consistent fashion and did not depend upon factors other than patients’ established clinical case records. An earlier, smaller study from northeast England of 117 cases of PBC diagnosed during 1966-1979 had shown evidence of seasonality in symptoms, particularly in the spring and early summer.10 However, this older analysis was based on a relatively small number of cases and did not use as rigorous methods to define date of diagnosis. Nevertheless, our new findings are entirely consistent with the older study and provide robust evidence for a seasonal effect, with a peak in the month of June. Table 4. Numbers of Symptomatic and Asymptomatic Patients by Month of Diagnosis Month

Numbers of Symptomatic Patients

Numbers of Asymptomatic Patients


25 29 26 37 34 36 31 35 27 27 24 24 355

27 31 35 33 37 58 40 24 42 39 50 34 450

Fig. 2. Plot of raw data and expected numbers in asymptomatic cases of PBC.

The findings are very supportive of our prior hypothesis that a primary factor influencing temporal heterogeneity of PBC is related to exposure to a seasonally varying environmental agent occurring close to diagnosis or at similar times before diagnosis. This is, at first sight, very surprising, because many studies have demonstrated that there can be a long latency between the development of AMA positivity in an individual and the presentation of overt disease. However, it should be noted that the June peak has arisen because of an excess of approximately 31 cases over the expected number. As we understand more of the possible cause of PBC, it is clear that both the etiology and clinical course may be influenced by a range of genetic and possible environmental factors. Hence, we can hypothesize that this seasonality shown here may have arisen in a subset of individuals who are genetically or otherwise predisposed to the effect of a seasonally varying environmental agent with a very short latency period. The findings are also supported by the previous revelation of space-time clustering in cases from the same dataset.5 It is also of interest that patients in whom the diagnosis was made presumably nearer to disease onset (i.e., the early group) showed significant seasonal variation, whereas patients dying within 5 years of diagnosis—hence, presumably later in their disease and likely further from disease onset—showed no such seasonal variation (Table 3). We examined total clinic attendances and admissions to exclude the possibility that apparent seasonal variation in PBC diagnosis was merely reflecting overall numbers of office (i.e., clinic) attendances. Table 2 shows that this was not the case. In respect of the symptomatic at diagnosis group, though there was a significant sinusoidal variation in time of diagnosis, there was also a marked June peak in the asymptomatic at diagnosis group, accounting for 20 of the estimated 30 excess diagnoses in the

HEPATOLOGY, Vol. 54, No. 6, 2011

month. We cannot, however, completely exclude the possibility that seasonal variations in symptoms could have contributed to the overall seasonality in time of diagnosis (Table 4; Fig. 2). Seasonal variation in PBC is consistent with the involvement of at least one transient environmental agent in etiology. Examples of such factors that may be implicated include infections, air pollution, and diet. Putative infectious agents include E. coli, Novosphingobium aromaticovorans, and human beta retrovirus.6-9,18-25 Although infections appear to be the most plausible explanation, other possibilities should not be dismissed. Xenobiotics have also been implicated in the etiology of PBC. These commonly occur in pollutants, food preservatives, and pesticides.26 Atmospheric concentrations of pollutants are known to exhibit seasonal variation, providing support for a possible link with the onset of PBC.27,28 Diet often varies seasonally and may be associated with the onset of PBC. There are several possible mechanisms, such as consumption of fresh vegetables contaminated with pesticides or infection with a specific bacterium whose prevalence varies seasonally. In conclusion, this is the first study, in a large population using reliable methodology, to find seasonal variation among cases of PBC. There was a highly statistically significant excess of cases in the month of June, and the pattern exhibited a sinusoidal form of occurrence. These novel results provide further evidence for the role of one or more transient environmental agents in etiology, at least in some patients. Candidate agents include infections, atmospheric pollution, or dietary factors.

References 1. Kaplan MM, Gershwin ME. Primary biliary cirrhosis. N Engl J Med 2005;353:1261-1273. 2. Hirschfield GM, Liu X, Xu C, Lu Y, Xie G, Lu Y, et al. Primary biliary cirrhosis associated with HLA, IL 12A, and IL 12RB2 variants. N Engl J Med 2009;360:2544-2555. 3. Prince MI, Ducker SJ, James OF. Case-control studies of risk factors for primary biliary cirrhosis in two United Kingdom populations. Gut 2010;59:508-512. 4. Selmi C, Gershwin ME. The role of environmental factors in primary biliary cirrhosis. Trends Immunol 2009;30:415-420. 5. McNally RJ, Ducker S, James OF. Are transient environmental agents involved in the cause of primary biliary cirrhosis? Evidence from spacetime clustering analysis. HEPATOLOGY 2009;50:1169-1174. 6. Butler P, Valle F, Hamilton-Miller JM, Brumfitt W, Baum H, Burroughs AK. M2 mitochondrial antibodies and urinary rough mutant bacteria in patients with primary biliary cirrhosis and in patients with recurrent bacteriuria. J Hepatol 1993;17:408-414. 7. Selmi C, Balkwill DL, Invernizzi P, Ansari AA, Coppel RL, Podda M, et al. Patients with primary biliary cirrhosis react against a ubiquitous xenobiotic-matabolizing bacterium. HEPATOLOGY 2003;38:1250-1257. 8. Xu L, Shen Z, Guo L, Fodera B, Keogh A, Joplin R, et al. Does a betaretrovirus infection trigger primary biliary cirrhosis? Proc Natl Acad Sci U S A 2003;100:8454-8459.

MCNALLY ET AL.

2103

9. Haydon GH, Neuberger J. PBC: an infectious disease. Gut 2000;47: 586-588. 10. Hamlyn AN, Macklon AF, James O. Primary biliary cirrhosis: geographical clustering and symptomatic onset seasonality. Gut 1983;24:940-945. 11. James OF, Bhopal R, Howel D, Gray J, Burt AD, Metcalf JV. Primary biliary cirrhosis once rare, now common in the United Kingdom? HEPATOLOGY 1999;30:390-394. 12. Office for National Statistics. 2001 census: small area statistics and local base statistics [computer file]. ESRC/JISC Census Programme, Census Dissemination Unit, MIMAS, University of Manchester, Manchester, UK, 2001. 13. Metcalf JV, Bhopal RS, Gray J, Howel D, James OF. Incidence and prevalence of primary biliary cirrhosis in the city of Newcastle upon Tyne, England. Int J Epidemiol 1997;26:830-836. 14. Stolwijk AM, Straatman H, Zielhuis GA. Studying seasonality by using sine and cosine functions in regression analysis. J Epidemiol Community Health 1999;53:235-238. 15. Pearson K. On the criterion that a given system of deviations from the probable in the case of correlated system of variables is such that it can be reasonably supposed to have arisen from random sampling. Philosop Mag 1900;50:157-175. 16. Office of Population Censuses and Surveys Census Division, General Register Office (Scotland) Census Branch. 1981 census small area statistics: 100% population and households aggregated to ward level (Great Britain) [computer file]. Colchester, Essex, UK: UK Data Archive [distributor] SN, 1893, 1983. 17. Office for National Statistics. 1991 census: small area statistics and local base statistics [computer file]. ESRC/JISC Census Programme, Census Dissemination Unit, MIMAS, University of Manchester, Manchester, UK, 1991. 18. Gershwin ME, Selmi C, Worman HJ, Gold EB, Watnik M, Utts J, et al. Risk factors and comorbidities in primary biliary cirrhosis: a controlled interview-based study of 1032 patients. HEPATOLOGY 2005;42:1194-1202. 19. Butler P, Hamilton-Miller J, Baum H, Burroughs AK. Detection of M2 antibodies in patients with recurrent urinary tract infection using an ELISA and purified PBC specific antigens. Evidence for a molecular mimicry mechanism in the pathogenesis of primary biliary cirrhosis? Biochem Mol Biol Int 1995;35:473-485. 20. Tsuneyama K, Harada K, Kono N, Hiramatsu K, Zen Y, Sudo Y, et al. Scavenger cells with gram-positive bacterial lipoteichoic acid infiltrate around the damaged interlobular bile ducts of primary biliary cirrhosis. J Hepatol 2001;35:156-163. 21. Shimoda S, Nakamura M, Shigematsu H, Tanimoto H, Gushima T, Gershwin ME, et al. Mimicry peptides of human PDC-E2 163-176 peptide, the immunodominant T-cell epitope of primary biliary cirrhosis. HEPATOLOGY 2000;31:1212-1216. 22. Padgett KA, Selmi C, Kenny TP, Leung PS, Balkwill DL, Ansari AA, et al. Phylogenetic and immunological definition of four lipoylated proteins from Novosphingobium aromaticivorans, implications for primary biliary cirrhosis. J Autoimmun 2005;24:209-219. 23. Mattner J, Savage PB, Leung P, Oertelt SS, Wang V, Trivedi O, et al. Liver autoimmunity triggered by microbial activation of natural killer T cells. Cell Host Microbe 2008;3:304-315. 24. Kaplan MM. Novosphingobium aromaticivorans: a potential initiator of primary biliary cirrhosis. Am J Gastroenterol 2004;99:2147-2149. 25. Olafsson S, Gudjonsson H, Selmi C, Amano K, Invernizzi P, Podda M, et al. Antimitochondrial antibodies and reactivity to N. aromaticivorans proteins in Icelandic patients with primary biliary cirrhosis and their relatives. Am J Gastroenterol 2004;99:2143-2146. 26. Dronamraju D, Odin J, Bach N. Primary biliary cirrhosis: environmental risk factors. Dis Markers 2010;29:323-328. 27. Zabiegala B, Urbanowicz M, Namiesnik J, Gorecki T. Spatial and seasonal patterns of benzene, toluene, ethylbenzene, and xylenes in the Gdansk, Poland and surrounding areas determined using radiello passive samplers. J Environ Qual 2010;39:896-906. 28. Tiwari V, Hanai Y, Masunaga S. Ambient levels of volatile organic compounds in the vicinity of petrochemical industrial area of Yokohama, Japan. Air Qual Atmos Health 2010;3:65-75.