|Home | About | Journals | Submit | Contact Us | Français|
Meningioma, the most frequent tumor in the central nervous system, has few recognized risk factors. We explored the role of allergies in a population-based case-control consortium study of meningioma in five geographic areas. We also studied serum levels of a marker of atopic allergy (IgE) in a subset of study participants, a first for a study on meningioma. Participants (N = 1,065) with surgically resected, pathologically confirmed meningioma and controls (N = 634) selected via random-digit dialing were recruited and interviewed. Cases were less likely than controls to report history of physician-diagnosed allergy [odds ratio (OR) = 0.64; 95% confidence interval (95% CI): 0.51 – 0.80]. Also, cases (N = 295) had lower total serum IgE than controls [N = 192; OR = 0.85, 95% CI: 0.75–0.98 for each unit of Ln(IgE)]. Similar to glioma and cancers at several other sites, meningioma appears to have an inverse relationship with history of allergies and a biomarker of atopic allergy. Since some common opposing predisposition or developmental processes for allergy and meningioma may exist, further research into immune processes that can affect the incidence and natural history of meningioma is warranted.
Meningiomas are the most frequently diagnosed primary brain and central nervous system tumor accounting for 34% of brain tumors in the US (2002–2006) 1. Risk factors for meningioma remain relatively unexplored with ionizing radiation exposure to the head being the only established environmental risk factor 2, 3. Gliomas, the second most common subgroup of brain tumors, have been consistently inversely associated with history of allergy and allergy-associated IgE [reviewed in 4], indicating that allergy or related immune factors may be protective for glioma. Allergy history has also shown inverse associations with risk of cancers of the colon, larynx, esophagus, oral cavity, pancreas, stomach, uterine, cervix, and non-Hodgkin lymphoma 5, 6, suggesting that anti-cancer immune mechanisms related to allergy may be a prevalent phenomenon. However, allergy history has demonstrated a positive association with risk of bladder cancer, lung, prostate, lymphoma, and myeloma, although results have not been completely consistent across all studies 5.
The relationship between meningioma and allergies is to date preliminary but consistent: one case-control study, from the United Kingdom branch of the INTERPHONE study and including 475 cases and 1,716 controls, demonstrated significantly reduced risk with allergies [OR = 0.76; 95% CI: 0.61–0.96 7], while several others demonstrated a reduced but non-significant risk with odds ratios ranging from 0.87 to 0.98 8–11. For this report, we evaluated whether self-reported allergy history and total serum IgE, a biomarker of atopic allergy, were associated with meningioma case status in a multicenter case-control study. Compared to previous studies, this study has a larger size, population-based recruitment method specifically for meningioma, and a biomarker of atopic allergy.
Eligible cases included all persons aged 20 to 79 with newly diagnosed, pathologically confirmed, intra-cranial meningioma among residents of the states of Connecticut, Massachusetts, and North Carolina as well as the Alameda, San Francisco, Contra Costa, Marin, San Mateo, and Santa Clara counties of California, and the Brazoria, Fort Bend, Harris, Montgomery, Chambers, Galveston, Liberty, and Waller Counties of Texas, from May 1, 2006 to December 12, 2009. Cases were identified through the Rapid Case Ascertainment (RCA) systems at most sites (Connecticut, Massachusetts, North Carolina, and California), and through review of hospital pathology departments and statistical-based tumor registries at sites without a formal RCA mechanism (Texas). Controls were selected by random-digit-dialing methods by an outside consulting firm (Kreider Research) and were frequency-matched to cases by five-year age interval (except for the youngest group which was 20–40 years), sex, and state of residence. Cases or controls with a previous history of meningioma and/or a brain lesion of unknown diagnosis were excluded. The study, consent forms and data collection instruments were approved by the Institutional Review Boards at the Yale University School of Medicine, Brigham and Women’s Hospital, the University of California at San Francisco, the M.D. Anderson Cancer Center, and the Duke University School of Medicine.
We contacted physicians of eligible cases in accordance with each site’s Institutional Review Board’s requirements to obtain approval to approach each case. Cases approved for contact by their physicians and controls identified by Kreider were sent a letter of introduction. Approximately 1–2 weeks after mailing the letters a trained interviewer contacted the potential study participant by telephone to administer the interview, or schedule it for a future date. Oral consent for study participation was obtained at the time of the telephone interview. Concurrently, written consents for specimen collection and medical release forms were also reviewed over the phone and participants were asked to sign, date, and mail forms to the study central office at Yale University prior to further contact for blood or saliva. The same interviewers conducted interviews for both cases and controls, and interviews averaged 43 minutes. The questionnaire included detailed questions about family history of cancer, reproductive factors including pregnancy and menstrual history, exogenous hormone history, demographics, medical history including history of allergy, and exposure to ionizing radiation, smoking, and alcohol consumption. Allergy was recorded as positive only if the study subject answered yes to “Did a doctor or other medical practitioner ever tell you that you had allergy?” Interviewers were instructed to probe the study participant further to ensure that a “yes” answer for allergy was confirmed by prescription medication and/or physician diagnosis. Additional questionnaire information used for the current analysis included basic demographic information, and history of smoking, asthma, and eczema. A licensed phlebotomist obtained blood and/or saliva specimens for laboratory analysis at a home visit, which occurred subsequent to the interview.
We identified 1,990 eligible cases and 1,652 eligible controls for this analysis. Ninety-three percent of eligible cases had a consenting physician. Among cases with a consenting physician, 60% participated in the interview portion of the study, while 54% of eligible controls participated in the interview. Allergy history was therefore available for 1065 cases and 634 controls. Ninety percent of interviewed cases and 74% of interviewed controls agreed to provide a blood or saliva specimen. Blood samples for IgE testing were available for 295 cases and 192 controls.
Total IgE was assessed using Phadia Diagnostics ImmunoCAP assay (Portage, MI) as previously described 12. Briefly, 40 uL of serum were incubated on the mix of allergens or anti-IgE antibodies bound to solid-phase ImmunoCAP. Incubations with enzyme-labeled antibodies against the heavy chain (constant) for total IgE were followed by incubations of developer and stop solutions, and measurements were made on a UniCAP 100. IgE was analyzed as a continuous variable, and also as a categorical variable using clinically-accepted cutoffs: IgE levels of 100 kilounits/liter (kU/L) are clinically “elevated,” IgE levels of 25 to 100 (kU/L) are “borderline,” and IgE levels of <25 (kU/L) are “normal.”
For the current analysis, we constructed two datasets. First, we analyzed the subset of cases and controls with available blood for IgE testing. Second, we used our full set of available cases and controls with questionnaire data to ensure that the subset with serologic testing was equivalent to the larger series, and also to gain better power for questionnaire-derived variables. Descriptive statistics including T-tests and chi-square tests were used to examine the association between the risk of meningioma and independent covariates. Unconditional logistic regression was used to provide maximum likelihood estimates of odds ratios (ORs) (adjusted and unadjusted) with 95% confidence intervals (CIs) using the statistical package PC-SAS 13. For all analyses, as indicated in the tables, odds ratios were adjusted for age (continuous), smoking (2 levels, <100, and ≥100 cigarettes or more in a lifetime), gender, and race (white/non-white). We also adjusted odds ratios by a measure of socioeconomic status (SES). Approximately 16% of our study population did not provide income information. Education level (4 levels) and income (5 levels) were correlated (Spearman’s r = 0.33, P < 0.0001), and we chose to adjust on education level only (4 levels). Concordance and kappa measures were also calculated in SAS.
Cases and controls were well-matched on age and ethnicity, and those from the subset with blood collection were not significantly different from the larger dataset with regard to age, gender, smoking, or any of the variables of interest for this analysis, i.e., prevalence of allergy, asthma, eczema, or smoking (Table 1). We had 1.6 times as many cases as controls because of the delayed recruitment of controls in our study relative to cases necessary to maintain our matching scheme. Cases were less likely than controls to report history of health provider-diagnosed allergy (23% of cases versus 33% of the controls, P < 0.0001, Table 1). Allergies were also related to demographic variables including income and education (Table 1).
Age, gender, smoking, race, and education-adjusted odds ratios for allergies for cases compared to controls was 0.64 (95% CI: 0.51–0.80, P<0.0001, Table 2). Cases also were less likely than controls to report histories of asthma (adjusted OR = 0.65; 95% CI: 0.50–0.86, P = 0.003), but not eczema (Table 2). Odds ratios for these same variables were consistent among the subset of subjects from which blood was obtained (OR = 0.64, 0.51, and 0.66 for allergies, asthma, and chicken pox, see Table 2); both allergy and asthma were significant (P = 0.03 and 0.01, respectively).
IgE values were natural log (Ln) transformed to approximate a normal distribution (as in reference 12). Mean IgE levels were different between cases and controls (P = 0.02, Table 3). In adjusted analyses, IgE was lower among cases compared to controls when assessed as a continuous variable [OR = 0.85 for each unit Ln(IgE); 95% CI: 0.75–0.98], and in clinically-relevant categories (Table 2); however, statistical significance was not reached for the latter. When split into quartiles, the lowest quartile of IgE (0 to 9.01 kU/L) had a gender, race, smoking, age, and education adjusted OR = 0.64 (95 % CI: 0.37–1.13, P = 0.01) when compared to the highest quartile (>66.65 kU/L) as a reference (data not presented in Tables). Mean levels of Ln(IgE) were 17% lower in female cases than male cases (P = 0.003) but only 10% lower in female controls compared to male controls (P = 0.13). When stratified by gender, the odds of reporting allergies was lower in meninigoma cases in both females and males (OR = 0.62 and 0.69, respectively, in the full dataset). However, IgE was only significantly lower in female cases compared to female controls (for females, adjusted OR = 0.80 for each unit Ln(IgE), 95% CI: 0.67–0.95, P = 0.01; for males, adjusted OR = 0.98 for each unit Ln(IgE), 95% CI: 0.78–1.22, P = 0.86, data not presented in Tables).
History of clinician-diagnosed allergy and IgE levels displayed limited concordance with each other, with Kappa values of less than 0.1 in both cases and controls (Table 4). Because of this, we evaluated whether reported allergies and IgE levels were reflecting different phenomena. We stratified the data by using study subjects that did not report allergy and with normal IgE as a control, we calculated odds ratios for those without allergies but with borderline/high IgE, and for the normal and borderline/high IgE groups with allergies (Table 5). The odds of being a case was about two-fold lower for those with allergies, and IgE levels did little to further reduce the magnitude of the odds ratio (i.e., comparing the OR of 0.53 to 0.54, Table 5).
Our data indicate that allergies, or associated immune phenomena, are less frequent among individuals who are diagnosed with meningioma, compared to controls. Serum IgE levels, a biological marker of atopic allergy, are also lower, providing support to the questionnaire-derived data. The association for IgE appeared to be stronger for females than males. Self-reported physician-diagnosed allergies and IgE were poorly concordant with each other, suggesting that they may be indirectly related to each other via an immunomodulatory mechanism that protects from meningioma and promotes allergy.
Several mechanisms have been proposed to explain the link between allergy and cancer, one or more of which may apply to meningioma. The characterization of such a mechanism may help to identify those persons most susceptible for the disease, promote the development of early detection methods, and/or assist in the development of anti-meningioma therapies based on immune rejection of the tumor. One mechanism is commonly called “immune surveillance” which posits that active immune systems that are highly allergic may also be more competent in recognizing and responding against nascent tumors which are recognized as foreign 14. Other theories posit a more specific mechanism related to allergy, such as the promotion of an active immune rejection of the tumor based on activation of macrophages, mast cells, and eosinophils, which are characteristics of allergy 15. Some cancers and in particular intracranial gliomas, are highly adept at suppressing specific cell-mediated anti-tumor immune mechanisms via the expression of immunosuppressive cytokines, and decoy and death receptors 16–18. Our current data do not illuminate on particular mechanisms for the role of allergies in meningioma but illustrate this as a potentially fruitful future research area.
One question that may arise is the accessibility of meningioma to the immune system. Meninges exist only partially behind the blood brain barrier, as significant permeability exists to contrast agents. Meningioma tumors themselves are highly vascular and completely lack capillaries capable of establishing the blood-brain barrier, and are supplied predominantly by the dural vessels 19–21. Both nascent meningeal neoplasms and the mature tumors are therefore far more exposed to the normal immune repertoire, in contrast to gliomas and other CNS tumors which must form, in their early stages, within the restricted environment behind the blood brain barrier. Meningioma, with a relatively benign and differentiated phenotype compared to glioblastoma, may evade immune recognition via the fact that tissue phenotype more closely approximates normal arachnoid fibroid tissue, and the encapsulated tumor does not elicit cellular invasion or destruction. Meningiomas do, however, express autoantigens that can be recognized as foreign tissue to the immune system 22, 23. This observation, combined with our current results (significant deficit of reported allergy and biologically measured IgE among cases), suggest that immune recognition of the tumor combined with appropriate activation may be helpful in treatment and prevention modalities for meningioma, and aspects of immune recognition of meningioma may provide suitable targets for early detection.
IgE levels were found to be related to several demographic/behavior variables in the expected directions – including gender (higher in males), ethnicity (lower in whites), and smoking (higher in smokers). We note that gender differences in IgE levels were more extreme among cases (17% difference, P = 0.003) than controls (10% difference, P = 0.13). The IgE assay is highly quantitative with an intra-class correlation coefficient of 0.99 in our laboratory 24, so this difference is likely to be genuine. Females exhibited a strong case-control difference with IgE (OR = 0.80, P = 0.01), while males were largely null (OR = 0.98). Whether gender-specific immune factors can account for some of the large 2:1 gender bias in meningioma rates 3 cannot be answered with these data, but is an intriguing point for further study. As allergies are more common in female adults than males, with possible influence from hormones 25, it is not likely that allergy plays a role effecting the absolute rate difference between the genders.
Several epidemiology studies have examined meningioma and the role of allergies. The meningiomas were often collected with other central nervous system tumors including glioma. Nearly all of these studies showed reduced (though not statistically significant) odds ratios indicating an inverse relationship consistent with the data in the current report 8–11, 26. Several prior studies had an issue with sufficient power (small sample size) but in one study with a relatively large sample (N = 1,201), only eczema was significantly inversely related to meningioma status (OR = 0.74; 95% CI: 0.60–0.91) but not overall allergy (OR = 0.87, 95% CI: 0.66–1.44) 11. While eczema was not significantly related to meningioma in our study, it is interesting to note that eczema is typically mediated by non-atopic (ie., non-IgE mediated) mechanisms. This result in Wigertz et al. 11, agrees closely with another large multinational study (N = 319 meningiomas) that demonstrated an OR of 0.89, 95% CI: 0.65–1.22 8. The definition of “allergy” is changeable in various questionnaire instruments: most require allergy diagnosis to be physician defined, resulting in prevalence of around 20–40% 8. Self-reported, self-diagnosed allergies can demonstrate prevalence up to 85% 27. Our current analysis required diagnosis by a doctor or health practitioner, and we found that the prevalence of allergies was 33% among controls. One prior study showed an overall significant result with regards to allergy and meningioma (OR = 0.76; 95% CI: 0.61 – 0.96) 7. Although this study from the United Kingdom combining two catchment regions did not specify “physician-diagnosed” allergy, the allergy prevalence of 38% among controls was similar to ours. Interestingly, Schoemaker et al., captured data on age of allergy onset and found that childhood onset allergies exhibited the most risk reduction with odds ratios of 0.43, 0.50, and 0.46 for asthma, hay fever, and eczema, respectively 7. The fact that childhood-onset allergies are more likely atopic than adult allergies is not entirely consistent with the results in the current analysis which points to the most risk reduction for non-atopic allergy. Further research is necessary to understand the finer points regarding mechanism.
Our study is subject to several potential sources of bias or measurement error, which may have affected our results. Because our cases are obtained largely via population-based registries, and there is virtually no early mortality, ascertainment bias of cases is not likely a problem. In addition, the subset of subjects that provided blood was highly similar to the larger set of subjects (both cases and controls, Table 1), bolstering confidence in the IgE results. Since serum levels were obtained after diagnosis and treatment of meningioma, it is possible that “reverse causality” played a role – i.e., that the disease itself or treatment modalities induced suppression of IgE levels. We consider this unlikely for several reasons – first, meningiomas unlike gliomas are not noted for immunosuppressive characteristics. IgE measurements are only taken at one time point and are not likely to adequately represent a life history of atopic allergy. This may be one reason for the poor concordance of self-reported, physician-diagnosed allergy and IgE. Also, treatment modalities for meningioma are not known to affect immune parameters; all of our cases have had surgical intervention, but none were subjected to radiation or cytotoxic chemotherapy. Bias in information can also result from control selection. Controls recruited via random-digit dialing tend to be more highly educated, have higher income, and therefore higher socioeconomic status than cases recruited from the same population. There was some evidence for this phenomenon in the current study regarding income and education status (Table 1), however, statistical adjustments for income did not appreciably change odds ratios for allergy and IgE in relation to meningioma (data not shown). Finally, there is a potential for recall bias. Since it is unlikely for the public to have any prior knowledge of allergies being related to meningioma, differential recall between cases and controls is unlikely. Also we note that many meningioma patients typically recover from surgery with full mental faculties, and lack of recall due to the disease is unlikely. Some lack of recall is possible in both cases and controls, which will likely be non-directional, and therefore bias results toward the null. The results are therefore likely to be robust despite potential sources of bias.
This work was supported by the Brain Science Foundation, the Meningioma Mommas and by NIH R01 grants CA109468, CA109461, CA109745, CA108473, and CA109475.
We thank Sydnee Crankshaw, MPA, Katherine Wagenman, Joe Patoka, and Lisa Padilla for sample management; and Koren Jones, MPH, Katherine Saunders, MS, and Estella Kanevsky, MPH for project management expertise and Donna Dello Iacono, NP, PhD for assistance with clinical expertise. We also thank the participants of this study, and the physicians of cases.