Monosodium glutamate sensitivity in asthma☆☆☆★

Article Outline

• Abstract
• METHODS
  • Recruitment of patients
  • Initial screening
  • Study protocol
  • Statistical analysis
• RESULTS
  • Enrollment of subjects
  • Screening placebo challenge
• DISCUSSION
• Acknowledgements
• References
• Copyright

Abstract 

Background: Questions have been raised since the early 1980s about monosodium glutamate (MSG) and provocation of asthma. Because MSG is widely available as a chemical in both natural foods and as an additive in many prepared foods, the need to define the relationship of MSG to asthma is of great importance. Objective: The purpose of this study was to determine whether MSG ingestion induces asthma attacks in asthmatic subjects. Methods: With single-blind, placebo-controlled screening challenges, 100 subjects with asthma (30 subjects with a history of Oriental restaurant asthma attacks; 70 subjects with a negative history) were challenged with 2.5 g of MSG. A total of 78 patients were proved to have aspirin-sensitive asthma. Results: No patient had a significant fall in FEV₁ value or the development of asthma symptoms during the MSG challenge. The mean change in FEV₁ with MSG challenge was no different from that of placebo challenge. Subjects with an MSG-positive history showed no significant percent decrease in FEV₁ values after placebo challenges compared with MSG 2.5 g oral challenge (P = .28). In the group with an MSG-negative history, there was no statistical difference in the change in lowest FEV₁ values between the placebo and MSG challenges (P = .44). The exact 1-sided 95% confidence interval (CI) for the probability of MSG sensitivity in individuals with aspirin-sensitive asthma (negative history) is 0% to 0.04%. When combined with previous studies that did not demonstrate MSG-provoked asthma, the 95% CI is 0% to 0.03%. For patients with an MSG-positive history, the exact 1-sided 95% CI for the probability of MSG sensitivity in this study was 0% to 0.07%, which is somewhat wider because of the smaller sample size. Conclusions: MSG challenges in subjects with and without a perceived sensitivity to MSG failed to induce signs or symptoms of asthma. Therefore in view of the poorly conducted studies that proposed that MSG induced asthma and the subsequent studies that failed to confirm those findings, it is important to maintain a healthy skepticism about the existence of MSG sensitivity in individuals with asthma. (J Allergy Clin Immunol 1999;104:305-10.)

Keywords:  Monosodium glutamate, asthma, aspirin-sensitive asthma, single-blind, placebo-controlled oral challenges

Abbreviations:  CI , Confidence interval, GCRC , General Clinical Research Center, MSG , Monosodium glutamate, NSAID , Nonsteroidal anti-inflammatory drug, PEFR , Peak expiratory flow rate

A possible association between monosodium glutamate (MSG) and asthma was first proposed in a letter to the editor from Allen and Baker in 1981.¹ They presented 2 patients with a history of asthma attacks occurring 12 hours after the ingestion of Chinese restaurant meals. Both patients experienced bronchospasm, with declines in peak expiratory flow rates (PEFRs) 10 and 12 hours after the ingestion of 2.5 g of MSG. Allen et al² expanded their initial investigation to include an additional 30 patients. Of 32 patients, 14 patients were said to have asthmatic reactions as a result of ingesting 2.5 g of MSG, with elapse times from ingestion of MSG to onset of 20% decline in PEFR ranging from 1 to 12 hours. Included in this group were 14 subjects with a history of asthma attacks during Chinese restaurant meals and 18 subjects with aspirin/nonsteroidal anti-inflammatory drug (NSAID) sensitivity (subpopulations of asthmatic subjects cited by Allen et al²), who were at increased risk for asthmatic reactions from the ingestion of MSG.

In 1987, the study by Allen et al² was of great interest, particularly in view of MSG’s availability as a chemical in both natural foods and as an additive in many prepared foods. Normal dietary intake of MSG is approximately 1 g per day in its free form, with an additional 0.55 g per day coming from added MSG.³ Some foods (eg, tomatoes, Parmesan cheese, and soy sauce) contain naturally high levels of MSG.⁴ Therefore exposure to MSG in all asthmatic patients worldwide is considerable, and a potentially intriguing explanation for ongoing asthma might be the daily ingestion of MSG.

However, the interpretation of the study by Allen et al² is complicated by multiple problems in the design and implementation of the study. These include the use of effort-dependent flow meters instead of the more reliable method of flow volume measurement, withdrawal of theophylline on the first day of study, administration of β-agonists to some unnamed study participants in the morning if baseline PEFR values were low, and MSG challenge in some patients at night with others challenged in the morning. In addition, the reported results of decline in PEFRs between 1 and 12 hours does not suggest a reaction to a challenge substance with a specified elapsed time, but rather appears to be a reflection of spontaneous variation in lung function. Subsequent studies have attempted to clarify the issue as to whether MSG induces bronchospasm in individuals with asthma. Schwartzstein et al⁵ reported that MSG at 1.5 g (1 g less than used by Allen et al²) was not shown to induce bronchospasm in the 4 hours after outpatient oral challenges in 12 patients. Germano et al⁶ found that 1 of 30 patients with asthma who underwent outpatient single-blind screening oral challenges with MSG up to 6 g (7.5 g cumulative dose over 2 hours) had a significant reduction in FEV₁ values. However, when the 1 preliminary “reactor” patient was rechallenged with the “reactor” dose of MSG under double-blind placebo-controlled conditions, the response to MSG challenge was then found to be negative. Both of these studies were further limited by the fact that only 1 patient in the study of Schwartzstein et al⁵ and 2 patients in the study of Germano et al⁶ had a history of bronchospasm after an MSG-laden meal. Moneret-Vautrin⁷ reported that 2 of 30 asthmatic patients, for whom corticosteroids and theophylline were withheld for weeks before the study, were reported to have experienced the development of bronchospastic reactions after oral challenges with 2.5 g of MSG. Because the 2 patients also experienced significant declines in PEFRs during placebo challenges, the 15% changes in PEFRs during MSG challenges were not interpretable. Recently, Woods et al⁸ published an outpatient study in which 12 subjects with a perceived history of MSG-provoked asthma underwent a randomized, double-blind, placebo-controlled MSG challenge protocol. The authors were unable to demonstrate MSG-provoked asthma attacks, changes in bronchial hyperresponsiveness, or soluble markers of inflammation.

In light of the limited information available and conflicting opinions in the literature regarding MSG and asthma, we initiated an inpatient study to determine whether MSG ingestion induces asthma attacks in asthmatic subjects.

Back to Article Outline

METHODS 

Recruitment of patients 

From our practices, referring physicians, and newspaper advertisements, 2 groups of asthmatic patients were recruited. Group A patients had experienced asthma attacks in Oriental restaurants, believed that MSG induced their asthma attacks, and were attempting to avoid MSG. Group B patients had suspected aspirin-sensitive asthma, were referred to the Scripps Clinic for aspirin desensitization, and did not have a perceived sensitivity to MSG.

Initial screening 

Group A and B patient volunteers underwent an initial telephone screening interview by the nurse coordinator or study physician to confirm a history of asthma and attacks of asthma associated with ingestion of food and drink in Oriental restaurants and/or aspirin/NSAIDs. If history was suggestive, patients were scheduled for an initial visit where the diagnosis of asthma was established by either a 20% improvement in FEV₁ values after β-agonist inhalation or a positive methacholine inhalation challenge. If copies of medical records from referring physicians were available and reversible obstructive airway disease had been documented previously, further documentation was not repeated. Prior asthma attacks induced by aspirin/NSAIDs and requiring emergency department treatment or hospitalization were also accepted as evidence for the diagnosis of asthma.

Usual medications for maintenance of asthma remissions were continued. If a patient was having an exacerbation of asthma from sinusitis, viral respiratory illness, or other causes, such events were treated and cleared before admission to the general clinical research center (GCRC). If the FEV₁ values were more than 70% predicted without requiring a β-agonist by inhalation to achieve this result, the patient underwent screening oral placebo challenge studies. If FEV₁ was less than 70% of predicted value, prednisone (30 to 40 mg/day) was initiated and 1 or more weeks later the patient’s condition was re-evaluated for potential admission to GCRC.

Challenge capsule preparation. MSG (Ajinomoto USA, Inc, Teaneck, NJ) was purchased locally (Vons, San Diego, Calif) by a GCRC nurse coordinator. Gelatin capsules were filled by the Green Hospital pharmacy with 500 mg of MSG or 500 mg of sucrose (Spectrum Quality Products, Gardena, Calif). MSG (2.5 g; 5 capsules) or sucrose (2.5 g; 5 capsules) were used as the challenge substances.

Spirometry. A wedge spirometer with integrated flow volume output (Koko Spirometer; Pulmonary Data Services Instrumentation, Inc, Louisville, Colo) was used for monitoring pulmonary functions and for the methacholine challenges used to establish the diagnosis of asthma. The protocol used for the methacholine challenges was derived from the manual provided by Pulmonary Data Services Instrumentation, Inc.

Criteria for a presumptive MSG-induced asthma attack. A 20% decline in FEV₁ values from baseline with or without accompanying symptoms was used as the criterion for a presumptive MSG-induced asthma attack.

Serum tryptase level determination. If a subject experienced a drop or more than 20% in FEV₁ value during MSG challenge, serum was collected for tryptase level determination at baseline, time of reaction, and 1, 2, and 3 hours after the drop in FEV₁ . Serum samples were assayed at the Virginia Commonwealth University in the laboratory of Lawrence Schwartz.

Study protocol 

Day 1: Admission to the GCRC. Subjects were admitted to the GCRC the day before the challenges. The patients underwent a complete history and physical examination, signed consent forms that had been approved by the Human Subjects Committee, and had maintenance medications verified and continued. Subjects who required maintenance long-acting bronchodilators to maintain airway function did not qualify for placebo challenge. A diet low in natural MSG was continued, and no MSG was added to their hospital diet. Subjects remained in the in-patient GCRC unit for the duration of the study.

Day 2: Single-blind screening placebo challenge. Placebo challenges were performed only if the FEV₁ baseline values were 70% or greater of their predicted value. Maintenance medications (if the subject was receiving medication therapy) were given at 6 AM ; breakfast was served between 6 and 7 AM , and 5 placebo capsules (500 mg sucrose each; single-blind) were ingested by the patient between 7 and 8 AM . FEV₁ values were measured every hour by the research nurse and at onset of symptoms for 12 hours. All symptoms and/or observed signs were recorded by the research nurse. A low-MSG lunch was served between 12 and 1 PM , and a second placebo challenge of 5 sucrose capsules was given between 1 and 2 PM followed again by hourly FEV₁ measurements. Failure to complete the screening challenge was defined as an FEV₁ change of more than 10% during the placebo challenge day.

Day 3: MSG challenge. Single-blind challenges with 2.5 g (5 capsules) of MSG were conducted if, during the baseline placebo challenge, FEV₁ values varied by less than 10% during the 12 hours of challenges. Furthermore, to assure day to day stability, baseline FEV₁ values on both days were required to be within 5% of each other. If these criteria were not met, the patient was classified as having unstable airways and did not undergo MSG challenge.

MSG challenges were identical to placebo with the substitution of MSG in the red capsules. Again, hourly FEV₁ measurements and recording of all signs and symptoms were performed for a total of 12 hours. At the 6-hour time point, 5 placebo capsules were given to conform to the challenge sequence on the preceding placebo challenge day.

If there were no significant changes in FEV₁ values during the single-blind screening challenges, the patient was either discharged from the GCRC or proceeded to additional oral challenge studies on subsequent days (usually aspirin).

If there was a 20% drop in FEV₁ value (presumptive reaction), serum tryptase levels were determined, and the patient underwent 2 double-blind placebo-controlled MSG challenges on days 4 and 5. Therefore days 4 and 5 would only apply to those patients who had a presumptive reaction to MSG during the single-blind MSG challenges. Two placebo and 1 MSG (2.5 g) challenges were conducted for each of the 2 double-blind challenges.

Statistical analysis 

Means, SEMs, 1-sided confidence intervals (CIs), and paired Student t test results compared the lowest declines in FEV₁ values during 12-hour challenges for placebo and MSG challenge days. The statistical software program used was StatView 4.01 (Abacus Concepts Inc, Berkeley, Calif) for Macintosh (Apple Computer, Inc, Cupertino, Calif) computers.

The method of calculation of exact 1-sided CIs for binomial probabilities was based on the following parameters: in the situation of no successes in the number (N) of independent trials with underlying probability (P ) of successes at each trial, the calculation is: the one-sided (1 – α)% CI for P is (0,P ), where P satisfies the equation (1 – P )ⁿ = α.⁹

Back to Article Outline

RESULTS 

Enrollment of subjects 

One hundred forty-two asthmatic subjects (38 subjects with a positive history for MSG reactions and 104 subjects with a negative history for MSG) responded to advertisements or were referred for studies. Fig 1 details the outcome of the 142 initial subjects.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1. 

  Flow chart of 142 patients who enrolled in the study.

One individual with a positive history believed that the study was too dangerous and declined our invitation to participate in the protocol. Seven other subjects could not arrange their schedules to accommodate the time commitment of the study. Of the 104 patients with a negative history, 13 patients were excluded because of daily dependence on salmeterol.

Screening placebo challenge 

Table I shows that 21 of 104 subjects did not successfully complete the screening/placebo challenges and were not challenged with MSG.

Table I. Patients excluded during placebo challenges

N , Nasal; p , pulmonary; BCM , beclomethasone; BDN , budesonide; TCN , triamcinolone; NT , not tested; FLN , flunisolide; FTN , fluticasone.

MSG challenges. One hundred patients, who met all of the study criteria, underwent challenges with 2.5 g of MSG.

Group A (MSG history positive). Table II outlines the placebo and MSG challenge data for the 30 subjects who reported sensitivity to MSG and who were actively avoiding MSG. There were no significant differences between percent changes in lowest recorded FEV₁ values after placebo challenge compared with percent changes in lowest recorded FEV₁ values after MSG oral challenge of 2.5 g in this patient population (P = .28, paired Student t test). Only 1 of 30 patients experienced a 20% decline in FEV₁ values during the single-blind screening challenge with MSG. She was without asthma symptoms throughout her MSG challenge; serum tryptase levels, determined at baseline, reaction, and 1, 2, and 3 hours after reaction were normal (<1 ng/mL). Subsequent double-blind placebo-controlled MSG challenges in replicate were negative, with the post-MSG (2.5 g) changes in FEV₁ values of less than 1%. The results of 1 of the 2 negative double-blind MSG challenge are provided in Table II. The exact 1-sided 95% CI for probability of MSG sensitivity in subjects with a perceived sensitivity to MSG (history positive) is 0% to 0.07%.

Table II. MSG challenge data in patients with a perceived sensitivity to MSG

n , Nasal; P pulmonary; Theo , theophylline; BDN , budesonide; HA , headache FLN , flunisolide; BCM , beclomethasone; DXM , dexamethasone; TCN , triamcinolone; FTN , fluticasone; TCM .

In Table II, information about concomitant medications, use of oral corticosteroids, and symptoms or signs experienced on the 2 days of challenges is given. Headaches occurred in 4 patients during placebo challenge and in 3 patients during MSG challenge.

Of the 30 subjects with a perceived sensitivity to MSG, 12 subjects were proved by challenge to have aspirin sensitivity. Baseline FEV₁ values were essentially the same when the subjects started their aspirin challenge the day after the MSG challenge.

Group B(no history of MSG sensitivity). Seventy subjects underwent screening placebo and MSG challenges without reaction to MSG. Eighty percent of challenged subjects had documented aspirin/NSAID sensitivity (56 subjects). The percent predicted FEV₁ mean value for the entire group was 87.9% (SEM, 1.77%). During the placebo challenge, the percent change to lowest recorded FEV₁ values ranged from 0% to 10% (mean, 3.13%; SEM, 0.66%). During the MSG challenges, the percent change to lowest recorded FEV₁ values ranged from 0% to 13% (mean, 3.67%; SEM, 0.74%). There was no statistical difference in the change in lowest FEV₁ values between the placebo and MSG challenges in this group (P = .44, Student paired t test). As also seen in Group A subjects, seemingly random symptoms occurred on both placebo and MSG challenge days. None of the patients experienced symptoms suggestive of asthma.

The exact 1-sided 95% CI for the probability of MSG sensitivity in subjects with aspirin-sensitive asthma without a history of asthma in Oriental restaurants is 0% to 0.04%.

Back to Article Outline

DISCUSSION 

Questions have been raised since the early 1980s about MSG and its provocation of asthma. Subsequent attempts to clarify the issue have been limited because of the small number of subjects studied, inadequate challenge protocols, use of effort-dependent measures of pulmonary function, and withdrawal of maintenance anti-asthmatic medications before the challenge with MSG. The elapsed time of up to 12 hours after ingestion of MSG and onset of asthma in the study of Allen et al² would mean that results from outpatient studies⁵, ⁶, ⁷, ⁸ were incomplete because of the fact that the subjects could not be followed for the required 12 hours. Therefore we studied 100 patients with asthma in an in-patient research unit using single-blind, placebo-controlled screening challenges, the same challenge dose of 2.5 g of MSG as Allen et al,² and failed to demonstrate any relationship between MSG and its provocation of bronchospasm.

The most efficient way to challenge patients with a substance not likely to induce asthma, particularly with large numbers of patients, is single-blind, placebo-controlled screening challenges.¹⁰ Furthermore, it is essential to ensure that such challenges are accurate. Therefore lung function, as a reflection of bronchial patency, must remain stable during an entire day of placebo challenges as a critical criteria before proceeding with the suspect substance during the days following the oral challenges. In the case of MSG, the work by Allen et al² claimed that MSG-provoked asthmatic reactions were delayed as long as 12 hours in some patients. To reasonably argue that the change in pulmonary function seen at 12 hours after MSG challenge was caused by MSG, stable lung function values must first be demonstrated for 12 hours. If a change in lung function (as the only measure of reaction) occurs during MSG oral challenges while the single-blind study is being conducted, such a change could be due to unstable airways or investigator or patient bias. Therefore any positive single-blind challenge result must be validated with 1 or more repeat double-blind, placebo-controlled challenges when the asthma is well controlled. This important step was not carried out by either Allen et al² or Moneret-Vautrin.⁷

The control of asthma with medications becomes an important issue when an attempt is made to determine whether a challenge substance can provoke an asthmatic reaction. A dilemma exists between the need to support bronchial patency with anti-asthmatic therapy during challenges, thus preventing spontaneous bronchoconstriction in the middle of a challenge, and the need to allow bronchoconstriction to occur if stimulated by the challenge substance. Ideally, all patients with asthma should be challenged with test substances when the asthma is in complete remission and they are not receiving anti-asthmatic medications. Such an ideal state is supplanted by the reality that the very asthmatic population at potential risk for reactions to a variety of ingested substances are those individuals with the least stable lung function and who require daily medications to maintain adequate lung function. We have shown that oral aspirin ingestion routinely induces bronchoconstriction in individuals with aspirin-sensitive asthma whose airway patency is maintained with corticosteroids and theophylline.¹¹, ¹², ¹³ Therefore some support of bronchial airways during challenges is critical because such measures prevent false-positive oral challenges. For our study, corticosteroids and theophylline were continued. Because medications were withdrawn in the Allen et al² and Moneret-Vautrin⁷ studies, some of the “positive” reactions to MSG seen in these 1987 studies most likely represented the effects of withdrawal of maintenance medications before the challenge, with expected concomitant decline in pulmonary function values. Furthermore, even with controller medications, 21 patients in our study (Table I) failed to qualify for MSG challenges because of unstable airways. These patients would have qualified for the Allen et al² and Moneret-Vautrin⁷ studies, thus casting further doubt on the validity of their reported results and conclusions.

With the use of stringent challenge protocols, we were unable to demonstrate a relationship between the ingestion of MSG and the provocation of asthma. Included in this study were 30 patients who felt that they had experienced asthmatic exacerbations as the result of the ingestion of MSG. Not only were these patients relieved to find out they were not MSG sensitive, but logical explanations for their Oriental restaurant–associated asthma attacks could be reasonably reconstructed for most of them. Ten patients had gastroesophageal reflux which, when treated with proton-pump inhibitors, allowed them to eat in Oriental restaurants without adverse events. Five subjects were allergic to foods that are readily available in Oriental restaurants. Twelve patients had anxiety and/or depression and admitted that the fear of MSG attacks was previously a significant problem in their lives. One prospective patient was so fearful of MSG that she thought she would die of an asthma attack during the challenges and therefore did not participate in the study.

Is there such an entity as MSG-induced asthma? Currently, the only evidence for this relationship are reported results from 2 poorly designed and implemented studies.², ⁷ According to Allen et al,² 71% of patients (10 of 14 patients) with a positive history of MSG-induced asthma and 22% of patients (4 of 18 patients) with aspirin-sensitive asthma had positive challenge responses with MSG (2.5 g). It is not possible to reconcile the Allen et al² study with our current study.

On the basis of our results, it is highly unlikely that the probability of MSG sensitivity in individuals with aspirin-sensitive asthma who have no history of Oriental restaurant asthma exceeds 4%. To date, including our current study, 45 patients with a positive MSG history have undergone negative challenge with MSG.⁵, ⁶, ⁸ The exact 1-sided 95% CI for the probability of MSG sensitivity in this group is 0% to 0.066%, which is somewhat wider than the previous intervals because of the smaller sample size. Nevertheless, in view of the poorly conducted studies², ⁷ (which proposed that MSG induces asthma in the first place) and the subsequent studies⁵, ⁶, ⁸ (which failed to confirm the Allen et al² results), we think it is important to maintain a healthy skepticism about the existence of MSG sensitivity in individuals with asthma.

Back to Article Outline

Acknowledgements 

We thank the research nurses in the GCRC for their important role in conducting this project, Susan Dastrup, RN, Nurse Coordinator for this study, and James A. Koziol, PhD, Biostatician for the GCRC, for his help in statistical analysis of our data.

Back to Article Outline

References 

1. Allen D, Baker G. Chinese-restaurant asthma. N Engl J Med. 1981;278:1154–1155
   • View In Article
2. Allen D, Delohery J, Baker G. Monosodium L-glutamate-induced asthma. J Allergy Clin Immunol. 1987;80:530–537
   • View In Article
   • MEDLINE
   • CrossRef
3. Filer L, Stegink L. A report of the proceedings of a MSG workshop held August 1991. Crit Rev Food Sci Nutr. 1994;34:159–174
   • View In Article
   • MEDLINE
   • CrossRef
4. Giacometti T. Free and bound glutamate in natural products. In:  Filer L,  Garattini S,  Kare M,  Reynolds W,  Wurtman R editor. Glutamic acid: advances in biochemistry and physiology. New York: Raven Press; 1979;p. 25–34
   • View In Article
5. Schwartzstein R, Kelleher M, Weinberger S, Weiss J, Drazen J. Airway effects of monosodium glutamate in subjects with chronic stable asthma. J Asthma. 1987;24:167–172
   • View In Article
   • MEDLINE
   • CrossRef
6. Germano P, Cohen S, Hahn B, Metcalfe D. An evaluation of clinical reactions to monosodium glutamate (MSG) in asthmatics, using a blinded placebo-controlled challenge [abstract]. J Allergy Clin Immunol. 1991;87:177
   • View In Article
   • CrossRef
7. Moneret-Vautrin D. Monosodium glutamate induced asthma: a study of the potential risk in 30 asthmatics and a review of the literature. Allergie et Immunologie. 1987;19:29–35
   • View In Article
   • MEDLINE
8. Woods R, Weiner J, Thien F, Abramson M, Walters E. The effects of monosodium glutamate in adults with asthma who perceive themselves to be monosodium glutamate-intolerant. J Allergy Clin Immunol. 1998;101:762–771
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (206 KB)
   • CrossRef
9. Lehmann EL. Testing statistical hypotheses. 2nd ed. New York: John Wiley; 1986;
   • View In Article
10. Stevenson DD, Simon RA, Lumry WA, Mathison DA. Adverse reactions to tartrazine. J Allergy Clin Immunol. 1986;78:182–191
    • View In Article
    • MEDLINE
    • CrossRef
11. Stevenson D. Oral challenges to detect aspirin and sulfite sensitivity in asthma. NER Allergy Proc. 1988;9:135–142
    • View In Article
12. Pleskow WW, Stevenson DD, Mathison DA, Simon RA, Schatz M, Zeiger RS. Aspirin-sensitive rhinosinusitis/asthma: spectrum of adverse reactions to aspirin. J Allergy Clin Immunol. 1983;71:574–579
    • View In Article
    • MEDLINE
    • CrossRef
13. Pleskow WW, Stevenson DD, Mathison DA, Simon RA, Schatz M, Zeiger RS. Aspirin desensitization in aspirin-sensitive asthmatic patients: clinical manifestations and characterization of the refractory period. J Allergy Clin Immunol. 1982;69:11–19
    • View In Article
    • MEDLINE
    • CrossRef