Autoimmune and Viral Risk Factors for Achalasia PDF

Received: 23 June 2021 | Revised: 1 October 2021 | Accepted: 30 November 2021 DOI: 10.1111/nmo.14312 ORIGINAL ARTICLE Autoimmune and viral risk factors are associated with achalasia: A case-­control study Charles E. Gaber1 | Cary C. Cotton2 | Swathi Eluri2 | Jennifer L. Lund1 | Timo...

Received: 23 June 2021 | Revised: 1 October 2021 | Accepted: 30 November 2021 DOI: 10.1111/nmo.14312 ORIGINAL ARTICLE Autoimmune and viral risk factors are associated with achalasia: A case-­control study Charles E. Gaber1 | Cary C. Cotton2 | Swathi Eluri2 | Jennifer L. Lund1 | Timothy M. Farrell3 | Evan S. Dellon2 1 Department of Epidemiology, Gillings School of Global Public Health, University Abstract of North Carolina at Chapel Hill, Chapel Background: Achalasia is a rare esophageal motility disorder of uncertain etiology. Hill, North Carolina, USA 2 Center for Esophageal Diseases and While past studies have indicated that autoimmune conditions and viral infections Swallowing, Center for Gastrointestinal may be associated with development of achalasia, these associations are yet to be Diseases and Biology, Division of Gastroenterology and Hepatology, examined in large, population-­based studies. Department of Medicine, School of Methods: A matched case-­control study was performed using administrative claim Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, data from the IBM MarketScan Commercial Claims and Encounters Database between USA 2000 and 2019. A history of selected autoimmune conditions and viral infections was 3 Department of Surgery, School of assessed using past medical claims. Multivariable conditional logistic regression was Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, used to account for the matched nature of the study design and further control for USA confounding by demographic and clinical characteristics when reporting adjusted Correspondence odds ratios (aORs). Evan S. Dellon, University of North Key Results: Among 6769 cases and 27,076 controls, presence of any of the autoim- Carolina School of Medicine, Bioinformatics Building CB # 7050, 130 mune conditions studied was associated with increased odds of achalasia (aOR = 1.26, Mason Farm Road, Chapel Hill, NC 27599, 95% CI: 1.11, 1.42). Scleroderma or systemic sclerosis (aOR = 8.13, 95% CI: 3.34, USA. Email: [email protected] 19.80) and Addison's disease (aOR = 3.83, 95% CI: 1.83, 8.04) had the strongest as- sociations with achalasia. Presence of any of the viral infections studied was also as- Funding information The project described was supported by sociated with an increased risk of achalasia (aOR = 1.58, 95% CI: 1.23, 2.01). Varicella the National Institutes of Health, through zoster virus (aOR = 3.84, 95% CI: 1.94, 7.62) and human papillomavirus (aOR = 1.77, Grant Award Number T32DK007634 (CEG and CCC) and P30DK034987. The content 95% CI: 1.15, 2.73) both had strong relationships with achalasia. is solely the responsibility of the authors Conclusions and Inferences: These findings suggest that achalasia may have autoim- and does not necessarily represent the official views of the NIH mune and viral components contributing to its etiology. Future mechanistic studies could target specific diseases and agents highlighted by this research. KEYWORDS autoimmune diseases, epidemiology, esophageal achalasia, esophagus, infections 1 | I NTRO D U C TI O N prevalence of 11 to 162 per 100,000 in the United States, with considerable increases in burden associated with advancing age.4-­6 Achalasia, a chronic motility disorder of the esophagus character- Symptoms of achalasia are generally severe and include dysphagia, ized by aperistalsis and loss of relaxation function of the lower es- regurgitation, heartburn, and weight loss, and the condition has ophageal sphincter, causes substantial morbidity.1-­3 Epidemiologic been associated with a substantial decrement to health-­related estimates report an incidence of 2 to 11 per 100,000 and a quality of life.7-­9 In addition to a high symptom burden, current Neurogastroenterology & Motility. 2022;34:e14312. wileyonlinelibrary.com/journal/nmo © 2021 John Wiley & Sons Ltd | 1 of 9 https://doi.org/10.1111/nmo.14312 2 of 9 | GABER et al. treatments for achalasia (botulinum toxin injection, pneumatic dilation, surgical myotomy, peroral endoscopic myotomy [POEM], Key points and esophagectomy) are invasive and expensive, may result in Prior studies suggest that autoimmune conditions and complications, and often require re-­intervention.10,11 Considering 4 viral infections may be associated with an increased risk the increasing incidence in older adults, high morbidity, and treat- of developing achalasia. ment burden, research into the causal mechanisms behind devel- In this matched case-­control study, we found that acha- opment of achalasia is critical. lasia cases were more likely than controls to have been Currently, the etiology of achalasia is uncertain. Given that neu- 12 previously diagnosed with autoimmune diseases and ral degeneration in the esophageal myenteric plexus, some have viral infections. postulated that an upstream viral insult coupled with an aberrant The associations detected in our study may inform the immune system response may be responsible for causing this dam- scope of future work examining the biologic mecha- age.13,14 This connection implicates infectious and autoimmune dis- 15-­17 nisms and pathogenesis of achalasia. eases as potential drivers of achalasia pathogenesis. Existing studies examining the association between these diseases with achalasia have demonstrated a positive relationship, though they are limited by small sample sizes and control selection that is not directly sampled from the underlying source population that gave rise to the Thus, they were excluded from the analysis to ensure the full cap- 18,19 cases. Such designs have high risk of bias. Larger epidemiologic ture of claims. The study was deemed exempt from ongoing review studies have been conducted but were reliant on self-­report data for by the Institutional Review Board at UNC-­Chapel Hill. past exposure to viral infections and autoimmune disease. 20 There is a need for large-­scale studies with well-­delineated source popu- lations and uniform exposure measurement to advance the under- 2.2 | Case ascertainment and control sampling standing of achalasia etiology. Therefore, we conducted a large case-­control study to quan- Incident achalasia cases were identified using the ICD-­9 (530.0) titatively estimate the associations between select autoimmune and ICD-­10 (K22.0) diagnosis codes for achalasia and CPT codes for conditions and viral infections with development of incident acha- achalasia treatments (Supplemental Materials). For our primary case lasia. Using nationwide employer-­sponsored private insurance ad- definition, we required at least one inpatient achalasia diagnosis or ministrative claim data from the United States, we sought to obtain two outpatient diagnoses on different dates (at least 30 days apart), precise association estimates with uniform exposure measure- consistent with our prior work.4 A single outpatient diagnosis did ment. These hypothesis-­generating insights will highlight in which not meet the case definition to protect against false positives from conditions future research on the etiology of achalasia may be most rule-­out codes potentially listed during a diagnostic evaluation. In fruitful. addition, all diagnosis codes had to be in the primary diagnosis posi- tion on the claim. This definition emphasizes specificity and positive predictive value to increase the likelihood that the cases included 2 | M ATE R I A L S A N D M E TH O DS are true cases. We conducted a sensitivity analysis of this case defi- nition that additionally required a procedure code for pneumatic di- 2.1 | Data source and study population lation, botulinum toxin injection, POEM (using the code for unlisted procedure of the esophagus as a proxy), myotomy, or esophagec- We conducted a retrospective case-­control study using administra- tomy within 180 days of diagnosis. tive claim data from 2001 to 2019 in the IBM MarketScan Commercial To be considered an incident case, a patient had to have their first Claims and Encounters Database (hereafter, “MarketScan”) to exam- claim containing an achalasia diagnosis in the patient's full medical ine the association between select autoimmune diseases and viral claim history, with at least 18 months of continuous prior insurance infections with the risk of incident esophageal achalasia. MarketScan enrollment during which there were no claims with a diagnosis code contains insurance enrollment and healthcare claim data on adults for achalasia. The earliest date listed on a case-­meeting sequence with commercial, employer-­sponsored insurance and their depend- of claims was defined as the index date. Controls were selected at 21 ents; these data cover over 39 million individuals. The study popu- the index date using incidence-­density sampling and matched on lation consisted of all enrollees ages 18–­6 4 with at least 18 months length of continuous insurance enrollment, age, and sex.22 In this of continuous medical insurance coverage, with no requirements re- control sampling design, controls can become future cases and are garding prescription drug coverage. This served as the source popu- only required to be disease-­free at the matched index date. 23 Four lation from which cases arose and matched controls were sampled. randomly selected eligible controls were required for each case. The While MarketScan contains data on adults 65 years of age and older, case-­control odds ratio (OR) in this design estimates the incidence their insurance may only be supplemental to Medicare coverage. rate ratio from the underlying cohort. 24 GABER et al. | 3 of 9 F I G U R E 1 Study design schematic depicting case ascertainment, control selection, and exposure assessment periods for four cases and their matched controls. The red cross symbol indicates when a case definition was met. Dark blue and light green sections of lines represent exposure assessment periods for cases and matched controls, respectively. Light blue and light green sections of lines represent the 6-­month buffer that was not used in assessing exposures for cases and matched controls, respectively. Dashed green line indicates control insurance-­ enrolled time that was not used in exposure assessment because it was during a period where the matched case was not enrolled 2.3 | Exposure assessment 2.4 | Covariate measurement Exposure to pre-­selected autoimmune conditions and viral infections Potential confounders included age, sex, time spent continu- was assessed by examining the claim history of each patient prior to ously enrolled in insurance, comorbidities, frailty, and degree the index date using ICD-­9 and ICD-­10 diagnosis codes. All available of healthcare utilization. Age was ascertained at the index date. past claims were used for exposure assessment, except for the recent Comorbidities were measured with the Gagne combined comorbid- time window six months prior to the index date. These six months were ity index using all available past claim history. The Gagne combined excluded from the exposure assessment period to reduce the chance comorbidity index combines the Elixhauser and Charlson comor- of differential exposure misclassification due to increased health- bidity indices and has been demonstrated to have improved predic- care utilization around the time of diagnosis during clinical workup of tive performance of patient mortality. 27 We similarly calculated the achalasia cases. A study schematic is provided in Figure 1 that pro- Kim claim-­based frailty index for each patient. 28 Though unlikely vides a visualization of the study design and variable measurement. to be a strong confounder in this younger cohort, frailty was in- Autoimmune conditions were selected based on a literature review cluded because it may serve as proxy for degree of healthcare uti- and included Grave's disease, Addison's disease, Crohn's disease, lization and subsequent likelihood of both exposure and achalasia Hashimoto’s disease, Sjogren's syndrome, Raynaud's syndrome, rheu- being documented via claims. All available past claims were used matoid arthritis, psoriatic arthritis, scleroderma or systemic sclerosis, in characterizing comorbidities and healthcare utilization, as this is psoriasis, sarcoidosis, lupus, ulcerative colitis, multiple sclerosis, uvei- known to increase the validity of the measurements compared to tis, celiac disease, alopecia areata, vitiligo, and type-­1 diabetes. Viral short lookback periods. Given that the comorbidities were meas- infections included cytomegalovirus, herpes simplex virus (HSV), vari- ured in the same window as exposure assessment, the assumption cella zoster virus (VZV), Epstein-­Barr virus, JC virus (JCV), human im- is that these covariates are not causal intermediaries affected by munodeficiency virus (HIV), human papillomavirus (HPV), and measles prior autoimmune disease. Healthcare utilization was measured as (all codes are listed in Supplemental Materials). These conditions were three separate variables: the number of hospitalizations, number considered to exist in a patient if one inpatient or two outpatient diag- of outpatient visits, and use of preventive services in the patients nosis codes (on separate dates) occurred in the medical claim history. claim history prior to the index date. A negative control exposure of superficial injury was used as a test of bias, wherein a null association should be expected between superficial injury and achalasia if confounding by healthcare utiliza- 2.5 | Statistical analysis tion is appropriately addressed. 25,26 Specific ICD diagnosis codes for superficial injury are listed in the supplementary table. Broadly, Descriptive statistics of patient demographics and clinical charac- superficial injury diagnosis codes are used to reimburse for medi- teristics were calculated for cases and controls. Multivariable con- cal encounters to treat skin cuts, wounds, and abrasions that do not ditional logistic regression was used to obtain adjusted ORs relating impact deeper layers of tissue. This negative control was selected exposure with achalasia development and account for the matched because we theorized that it will not be a cause of achalasia but may design used in control selection. A separate model was built for each still be subject to the same confounding factor of healthcare utiliza- exposure of interest. In each model, case status served as the out- tion, wherein a positive (but erroneous) relationship between super- come with predictor variables consisting of exposure status, geo- ficial injury and achalasia would be documented if this confounding graphic region, indicator variables for level of Gagne Comorbidity factor were not adequately handled. Index and Kim Frailty Index, number of inpatient visits in the past 4 of 9 | GABER et al. year, number of outpatient visits in the past year, and a dichotomous associations with limited precision were seen for type I diabetes, variable for use of preventive services. Adjusted odds ratios (aORs) multiple sclerosis, and Grave's disease. and 95% confidence intervals (CI) were reported for each exposure. Presence of any of the viral infections studied was associated All analyses were performed using SAS 9.4. with an increased odds of achalasia (Figure 3, aOR = 1.58, 95% CI: 1.23, 2.01). Among specific viral infections, VZV (aOR = 3.84, 95% CI: 1.94, 7.62) and HPV (aOR = 1.77, 95% CI: 1.15, 2.73) both had 3 | R E S U LT S strong relationships with achalasia. Measles and JCV also displayed positive relationships with increased odds of achalasia, though CIs 3.1 | Achalasia case and control characteristics were wide and meaningfully encompassed negative and null associ- ation values. No association was seen for HIV. The study population consisted of 6769 achalasia cases and Our negative control exposure, superficial injury, demonstrated 27,076 matched controls. Matching factors (age, sex, length of a null relationship with the outcome of achalasia in the adjusted enrollment) resulted in identical distributions of these variables model. The odds of developing incident achalasia among those with in cases and controls (Table 1). The median age of cases and con- a history of superficial injury were equal in those without such a trols was nearly 52 (IQR: 41, 59), and 52.2% of cases and controls history (aOR = 1.02, 95% CI: 0.91, 1.15). were females. The median length of enrollment prior to acha- A sensitivity analysis that re-­estimated these associations re- lasia diagnosis or control selection date was 3.4 years. Regarding stricted in a cohort of patients who were treated for achalasia within factors that were not matched on, cases had a higher burden of 180 days (Table 2) did not meaningfully change the estimates for the comorbidities (12% in the highest Combined Comorbidity Index composite exposure measure for autoimmune disease (aOR = 1.31, category compared to 6% of controls), were less likely to be in 95% CI: 1.06, 1.62). However, the association for viral infection was the most robust category of the frailty index (68.3% vs. 80.9%), attenuated (aOR = 1.12, 95% CI: 0.71, 1.76). Associations for specific had more extensive healthcare utilization prior to the index date, autoimmune diseases and viral infections are additionally presented and were slightly more likely to have received preventive services in Table 2. prior to the index date. Cases were slightly more likely to reside in the Northeast or South geographic regions compared to controls. There were 2359 (35%) cases treated with either pneumatic dila- 4 | DISCUSSION tion, myotomy, unlisted procedure of the esophagus (a proxy code for POEM), or an esophagectomy within 180 days of diagnosis. Achalasia is a debilitating esophageal disease, and its cause is not This treated subset of diagnosis-­b ased cases constitutes the sen- completely understood. Prior studies have suggested that viral in- sitivity analysis sample of cases. fections and autoimmune disease may play a role in the risk of de- veloping achalasia, but whether these associations persisted in large, population-­based datasets with rigorous procedures to mitigate bias 3.2 | Autoimmune and viral risk factors was unknown. We conducted a large case-­control study using ad- for achalasia ministrative claim data to examine differences in past exposure to autoimmune diseases and viral infections in achalasia patients com- The results of the multivariable conditional logistic regression pared to population-­based controls that were selected from the models are contained in Figures 2 and 3, presenting adjusted rela- underlying insurance-­enrolled source population that gave rise to tionships between autoimmune conditions (Figure 2) and viral infec- the cases. Of the conditions considered, both positive associations tions (Figure 3) with achalasia. Overall, adjusted associations varied of varying precision and null associations were found. Scleroderma across conditions from null relationships to increased risk associa- or systemic sclerosis was the autoimmune condition that had the tions. Presence of any of the autoimmune conditions studied was strongest association with achalasia, though positive associations associated with increased odds of achalasia (aOR = 1.26, 95% CI: were also seen with Addison's disease, sarcoidosis, Sjogren's syn- 1.11, 1.42). Among specific autoimmune conditions, scleroderma or drome, and inflammatory bowel disease. The strong association systemic sclerosis (aOR = 8.13, 95% CI: 3.34, 19.80) and Addison's seen with scleroderma or systemic sclerosis may be partly explained disease (aOR = 3.83, 95% CI: 1.83, 8.04) had the strongest asso- by the fact that achalasia and scleroderma both result in esopha- ciations with achalasia. Sarcoidosis (aOR = 1.91, 95% CI: 1.19, 3.05), geal motility abnormalities characterized by severely impaired bolus Crohn's disease (aOR = 1.69, 95% CI: 1.14, 2.50), and ulcerative transit. Thus, it is possible that some achalasia cases may have been colitis (aOR = 1.50, 95% CI: 1.01, 2.23) were also associated with previously misdiagnosed as scleroderma, contributing to the strong increased odds of developing achalasia. Positive associations with association demonstrated. With the development of high-­resolution limited precision (wide CIs) were seen for Raynaud's syndrome, manometry in the 2010s and the Chicago classification, it may have Sjogren's syndrome, alopecia areata, Hashimoto's disease, lupus, become easier to distinguish between these conditions in cases di- psoriatic arthritis, and uveitis. Null associations were seen for ce- agnosed after 2010. VZV was the viral condition with the strong- liac disease, psoriasis, rheumatoid arthritis, and vitiligo. Negative est association, but a positive association was also seen with HPV. GABER et al. | 5 of 9 TA B L E 1 Study population characteristics of achalasia and age, These hypothesis-­generating results may help refine the focus of sex, enrollment-­length matched controls, 2001–­2019 subsequent etiologic studies of achalasia. Incidence-­density Our findings were mostly harmonious with that of prior published Achalasia cases controls studies. Becker et al. demonstrated in a large case-­control study that N = 6769 N = 27,076 viral infections (OR = 4.07, 95% CI: 2.33, 7.10) and autoimmune con- Age, median (IQR) 51.6 (41.3, 58.5) 51.5 (41.3, 58.5) ditions (OR = 1.78, 95% CI: 1.27, 2.50) were both associated with Sex increased risk of achalasia. Through primary data collection and Male 3233 (47.8) 12,932 (47.8) requiring a gastroenterologist's diagnosis, the achalasia case defini- Female 3536 (52.2) 14,144 (52.2) tion in Becker et al. was stronger than our claim-­based definition, Region although exposure measurement relied on self-­report which may be Northeast 1266 (18.7) 4269 (15.8) subject to recall bias. Our finding of no association between psoria- North Central 1416 (20.9) 6380 (23.6) sis and achalasia (OR = 1.10, 95% CI: 0.83, 1.46) was divergent from South 2889 (42.7) 11,263 (41.6) Becker et al. who reported a positive association (OR = 2.70, 95% West 1133 (16.7) 4860 (18.0) CI: 1.34, 5.42). In a case-­control study of 114 cases and age and sex-­ Unknown 65 (1.0) 304 (1.1) matched controls with gastroesophageal reflux disease, Romero-­ Years enrolled prior to 3.4 (2.2, 5.7) 3.4 (2.2, 5.7) Hernández et al.19 found that achalasia patients were more likely to achalasia diagnosis, median (IQR) have an autoimmune disease documented in their medical records Achalasia treatments within 180 days of diagnosis (OR = 3.8, 95% CI: 1.47, 9.83). Booy et al. also found a positive re- Pneumatic dilation 362 (5.3) –­ lationship between autoimmune disease and achalasia (OR = 3.6, Myotomy 1829 (27.0) –­ 95% CI: 2.5, 5.3); however, the controls were derived external to the Unlisted procedure 188 (2.8) –­ study from the epidemiologic literature and did not represent the of esophagus source population which gave rise to the cases. To the extent that Esophagectomy 42 (0.6) –­ the controls from the literature have a different exposure level than No treatment 4410 (65.2) –­ the source population, the reported associations may be biased to- Gagne comorbidity score, n (%) ward or away from the null. −1 858 (12.7) 3853 (14.2) In particular, our finding of a strong positive association between 0 3295 (48.7) 15,758 (58.2) VZV infection and achalasia is congruent with evidence from prior 1 1266 (18.7) 4491 (16.6) studies that have examined this association. Nearly 30 years ago, 2 541 (8.0) 1489 (5.5) Robertson et al. found a significantly higher proportion of VZV ≥3 809 (12.0) 1485 (5.5) serum antibody titers in nine achalasia cases (33%) compared to 20 Kim Frailty Index, n (%) age and sex-­matched controls (0%).29 Most recently, Naik et al.30 Robust,