Estimating the Risk for Alpha-1 Antitrypsin Deficiency among COPD Patients: Evidence Supporting Targeted Screening

Abstract

Alpha-1 antitrypsin deficiency is known as a significant genetic risk factor for COPD for carriers of phenotype PIMZ, and for phenotypes PIZZ and PISZ. Genetic epidemiological studies for alpha-1 antitrypsin deficiency conducted by others on both COPD patients and concurrent non-COPD controls were used to estimate the risk factors for all six phenotypic classes (namely, the normal phenotype PIMM, and the 5 deficiency allele phenotypes: PIMS, PIMZ, PISS, PISZ, and PIZZ). Studies on alpha-1 antitrypsin deficiency in white (Caucasian) COPD and non-COPD populations in 6 countries were combined to obtain estimates of the prevalence of the PIS and PIZ deficiency alleles in the combined COPD and non-COPD cohorts. The odds ratios for each of the six phenotypic classes of alpha-1 antitrypsin deficiency were calculated for a hypothetical population of 19.3 million white COPD patients in the United States of America. This approach demonstrated that 1,829,673 alpha-1 antitrypsin deficiency patients would be detected by testing 19.3 million white COPD patients and 536,033 in white non-COPD concurrent controls. The odds ratios for each of the phenotypic classes among white COPD patients demonstrate highly significant decreases in the normal phenotype PIMM, no significant change in the PIMS and PISS deficiency phenotypes, but highly significant increases in the prevalences of the PIMZ, PISZ, and PIZZ deficiency phenotypes. The result of the present study supports the concept of targeted screening for alpha-1 antitrypsin deficiency in countries with large populations of white (Caucasian) COPD patients.

Key words: :

• COPD
• Alpha-1 Antitrypsin deficiency
• PI subtypes
• PI phenotypes
• Genetic epidemiology.

View correction statement:

ERRATUM

INTRODUCTION

In his comprehensive historical review of the discovery of AAT Deficiency by Carl-Bertil Laurell, Robin Carrell [[1]] stated that “In medicine, the plasma deficiency of α1-antitrypsin has drawn attention to protease-antiprotease imbalance as a contributory cause of chronic obstructive pulmonary disease.” Alpha-1-antitrypsin (AAT) is a 52-kDa α-1-glycoprotein composed of 394 amino acid residues and 3 asparagine-linked complex carbohydrate side chains [[11]]. It is produced mainly by hepatocytes and secreted into the blood where it acts as a circulating serine protease inhibitor whose principal substrate is neutrophil elastase (NE) [[12]]. In AAT Deficient individuals, some of the abnormal AAT is retained as pathologic polymers within the endoplasmic reticulum of the hepatocytes, resulting in a low plasma concentration [[13], [14]]. This deficit of AAT is usually insufficient to ensure lifetime protection of the lung from the proteolytic damage of NE, resulting in early onset of panlobular pulmonary emphysema in adults, especially in habitual tobacco smokers [[11], [15], [16], [17]]. In addition, this deficit of AAT also can result in the development of neonatal cholestasis that may progress to infant and juvenile cirrhosis. It also can result in slowly progressive liver disease in up to 40–50% of adults due to a pathologic aggregation of abnormal AAT polymers in hepatocytes [[18], [19]].

In a recent review [[20]] of the causes of death of persons with AAT Deficiency, it was concluded that severe AAT deficiency poses a significant threat to health, that severe airflow obstruction is a major determinant of mortality, and that liver and lung disease account for the excess mortality in affected individuals. These findings [[20]] support current efforts to enhance diagnostic recognition and treatment of AAT-deficient individuals. It is now well known that COPD patients are at increased risk for having AAT deficiency if they are a carrier of phenotype PIMZ [[2], [3], [4]] or are homozygotes or heterozygotes of phenotype PIZZ or PISZ [[5], [6]].

AAT Deficiency once was believed to be primarily a disease affecting whites from Northern Europe [[9]]. It is now known that AAT Deficiency is one of the most common serious hereditary disorders in the world, because it affects all major racial subgroups, and there are an estimated 120.5 million carriers and deficient subjects in the 58 countries surveyed worldwide [[10]]. The present study determined whether, in a comprehensive meta-analysis of the peer-reviewed medical literature, there were case-control studies that could be used to determine the prevalence of AAT Deficiency among COPD patients. In addition, these case-control studies have also been used to calculate odds ratios for COPD patients in each of the 6 phenotypic classes (PIMM, PIMS, PIMZ, PISS, PISZ, and PIZZ) of AAT Deficiency.

It has been proposed that as many as 24.9 million Americans are at risk for AAT Deficiency [[7]], including an estimated 11.2 million Americans who have already been diagnosed with COPD. < http://www.lungusa.org/diseases/ >. There are purported to be an additional 13.7 million adults with evidence of impaired lung function as yet not identified by physicians [[8]] suggesting that there could be as many as 24.9 million Americans at risk for COPD and hence AAT Deficiency. < http://www.nhlbi.nih.gov/health/public/lung/other/copd_fact.pdf >.

The objective of the present meta-analysis is to provide an estimate of the prevalence of AAT Deficiency in a hypothetical white COPD population and examine the effective utilization of the standards for diagnosis and management of individuals used by primary care and pulmonary physicians [[14]] in those countries with large white COPD populations.

We also have developed odds ratios to determine the relative risk for each of the six phenotypic classes: PIMM, PIMS, PIMZ, PISS, PISZ and PIZZ. It was also our intent with this exercise to develop a rationale for targeted testing for AAT Deficiency in those countries with large white COPD patient populations.

MATERIALS AND METHODS

We have used Hardy-Weinberg Equilibrium statistics in this meta-analysis to develop a more definitive estimate of the actual numbers at risk for AAT Deficiency in this hypothetical cohort of 19.3 million white COPD patients in the United States of America.

Sources of the case and concurrent control cohort data used in the present study

Criterion 1

Studies were rejected when the number of COPD patients was less than 100, when the control non-COPD populations were not concurrent, or where the phenotypic characterization, with regard to both the PIS and PIZ deficiency alleles, was incomplete. Only studies that tested for the normal phenotype (PIMM) and the 5 deficiency phenotypes (PIMS, PIMZ, PISS, PISZ and PIZZ) were selected.

Criterion 2

The papers used in the present meta-analysis were obtained through PubMed searches in the National Institute of Environmental Health Sciences/NIH Libraries and consist of a total of 10 studies on predominantly white populations in six countries as follows: 4 in the United States [[3], [21], [22], [23]], 2 in Germany [[24], [25]], and one each in Australia [[26]], Canada [[27]], Italy [[28]], and Sweden [[29]]. The populations selected from each of the five countries abroad were also judged to be representative of existing white populations in the United States, many countries in Europe, and those in Australia/New Zealand.

Criterion 3

Critical for a comprehensive Hardy-Weinberg Equilibrium statistical comparison of these 2 cohorts and to obtain 95% confidence intervals on all numerical estimates was the need to develop 2 cohorts of sufficient size. When the 10 studies listed here were combined, there are 3,144 COPD patients and 7,256 concurrent non-COPD controls.

Criterion 4

The United States Census 2004 < http://www.census.gov/statab/www/brief.html >. was used to estimate the white (Caucasian) population of 234 million out of the 291 million in the 2004 total United States population.

Criterion 5

In this estimated white population of 234 million in the United States, we have compared, as an exercise, the incidence of AAT Deficiency in an arbitrary estimate of 19.3 million COPD patients with 19.3 million concurrent non-COPD controls.

Criterion 6

The data from the individual cohorts for these 6 countries were weighted according to the total population of each country and combined to develop the COPD and non-COPD cohorts to get mean frequencies for the PIM, PIS, and PIZ alleles.

Criterion 7

These allele frequencies for each of the resulting COPD and non-COPD cohorts were used to calculate the total numbers of individuals in each of the 6 phenotypic classes of interest (PIMM, PIMS, PIMZ, PISS, PISZ, and PIZZ) and to develop odds ratios for each phenotypic class with 95% confidence intervals on all estimates using Hardy–Weinberg Equilibrium statistics.

Estimation of Statistical Reliability for Each Cohort Using Precision Factor Scores

To assess the statistical reliability of the COPD and non-COPD cohorts in the surveys for each country, a Precision Factor Score (PF Score) with a scale of 1 to 12 was developed by one of us (EF Bustillo), as described in an earlier publication [[7], [30]]. Since PF Score is inversely proportional to the values of the coefficient of variation (cv), which measures the dispersion of values in respect to the mean, the smaller value of cv the greater the value of the PF Score.

The cv depends on the total number of alleles (sample size), and on the allele frequencies of PIS and PIZ actually found. Thus, a low score is the result of small numbers of subjects in a given control cohort, or a series of control cohorts each with a small number of subjects, and low values for PIS and/or PIZ allele frequencies. A high score is given to those studies where the number of subjects is high or a series of control cohorts each with a large number of subjects, and high values for PIS and/or PIZ allele frequencies.

Estimating allele frequencies of PIM, PIS, and PIZ

The formulas developed by co-author Bustillo for developing estimates of the allele frequencies and 95% confidence intervals on all estimates of numbers and prevalence using Hardy-Weinberg equilibrium statistics were discussed in an earlier paper [[30]].

Odds Ratios

In the present paper, Dr. Bustillo has expanded the statistical analysis to include odds ratios for all phenotypic classes with 95% confidence intervals. The odds ratio [[31]] is a statistical method for comparing whether the probability of a certain event is the same for the COPD and non-COPD cohorts. An odds ratio of one implies that the event is equally likely in both cohorts, and an odds ratio greater than one implies that the event is more likely in one of the cohorts than the other cohort. An odds ratio less than one imply that the event is less likely in one of the cohorts than the other cohort. Odds ratios with 95% confidence intervals have been generated to compare the frequencies of each of the 6 phenotypic classes in the COPD and non-COPD cohorts.

RESULTS

Overview

Calculated statistical values of the PF Scores for the non-COPD control and COPD cohort samples from each of the six countries are shown in Table 1. These values range from 3.7 for the COPD sample from Sweden to 10.8 for the control sample for Australia.

Table 1 Calculated precision factor scores of the control and COPD cohorts in each of the 6 countries

		PF score	Cohort	Total sample	Statistical analysis of each cohort samples
Country		mean	no.	size	Mean	SD	Median	Range (Lower limit)	Range (Upper limit)
Australia	Control	10.8	1	1,303	1303	—	1,303	1,303	1,303
Australia	COPD	5.4	1	190	190	—	190	190	190
Canada	Control	6.4	1	723	723	—	723	723	723
Canada	COPD	4.0	1	163	163	—	163	163	163
Germany	Control	4.7	2	922	461	256	461	280	642
Germany	COPD	5.3	2	602	301	318	301	76	526
Italy	Control	4.1	1	490	490	—	490	490	490
Italy	COPD	4.1	1	260	260	—	260	260	260
Sweden	Control	4.4	1	619	610	—	619	619	619
Sweden	COPD	3.7	1	164	164	—	164	164	164
United States of America	Control	6.7	4	3,199	800	524	853	114.0	1,380
United States of America	COPD	6.6	4	1,765	441	388	343	114	965
Totals	Control	8.2	10	7,256	726	402	683	114	1,380
Totals	COPD	5.3	10	3,144	314	275	187	76	965

When the data for all 6 countries are combined, a PFS of 8.2 was obtained for the non-COPD control sample of 7,256, and 5.3 for the COPD sample of 3,144. These PF Scores indicate a high level of statistical reliability in the overall combined database of both the non-COPD and COPD cohorts.

Normal and Deficiency Allele Frequencies with 95% Confidence Intervals

The estimates for the frequencies of the normal allele PIM and the two deficiency alleles PIS and PIZ are given in Table 2. Comparison of the data for the non-COPD cohort with the COPD cohort shows that there is a significant decrease in the frequency of the PIM allele from 0.947 to 0.905, no change in the frequency of the PIS deficiency allele, and a significant increase in the frequency of the PIZ deficiency allele from 0.0141 to 0.0490.

Table 2 Normal and deficiency allele frequencies with 95% confidence intervals

			Allele frequency with 95% confidence intervals
Population surveyed	No.	Cohort size	PIM	95% CI	PIS	95% CI	PIZ	95% CI
AAT deficiency (non-COPD)	10	7,256	0.9467	0.9429–0.9502	0.0342	0.0313–0.0373	0.0141	0.0122–0.0161
AAT Deficiency (COPD)	10	3,144	0.9054	0.8978–0.91244	0.0379	0.0333–0.0429	0.0490	0.0438–0.0547

Data Summaries of PI Phenotypes for Normal and Carrier Phenotypes of the PIS and PIZ Deficiency Alleles

Comparison of the numbers for the normal PIMM phenotype (Table 3) in the COPD and non-COPD populations demonstrate a highly significant decrease with an odds ratio of 0.527 in the number of AAT Deficiency subjects from 17,296,164 in the non-COPD cohort to 15,820,296 in the COPD cohort. There is no significant change in the frequency of the PIMS phenotype, with an odds ratio of 1.063, in the number of AAT Deficiency subjects from 1,248,930 in the non-COPD cohort to 1,322,758 in the COPD cohort. In addition, this statistical analysis demonstrates a highly significant increase in the frequency of the PIMZ phenotype from 513,673 in the non-COPD cohort to 1,829,673 in the COPD cohort, with a highly significant odds ratio of 3.559.

Table 3 Data summaries of PI phenotypes for normal and carrier phenotypes with 95% CI in total populations of 19.3 million

Population surveyed	Total population	PIMM	95% CI	PIMS	95% CI	PIMZ	95% CI
AAT deficiency (non-COPD)	19,300,000	17,296,164	17,157,181–17,426,953	1,248,930	1,139,523-1,367,987	513,673	445,257–592,082
AAT deficiency (COPD)	19,300,000	15,820,296	15,556,976–16,068,195	1,322,758	1,155,186–1,512,361	1,711,894	1,519,363–1925,890
		PIMM	95% CI	PIMS	95% CI	PIMZ	95% CI
Odds ratios		0.527	0.526–0.527	1.063	1.061–0.066	3.559	3.548–3.571

Data Summaries of PI Phenotypes for the 3 Deficiency Allele Phenotypes and the Totals for Both Carriers and Deficiency Allele Phenotypes with 95% CI

The data for the PISS phenotypes (Table 4) in the non-COPD cohort and the COPD cohort demonstrate no significant increase (since the 95% confidence intervals overlap), with an odds ratio of 1.227, in the number of AAT Deficiency subjects from 22,546 to 27,649. In contrast, the data for the PISZ phenotype demonstrate a highly significant increase with an odds ratio of 3.869 in the number of AAT Deficiency subjects from 18,546 in the non-COPD cohort to 71,563 in the COPD cohort. The final comparison demonstrates a highly significant increase for the PIZZ phenotype, with an odds ratio of 12.168, in the number of AAT Deficiency subjects in the non-COPD cohort from 3,814 to 46,306 in the COPD cohort.

Table 4 Data summaries of PI phenotypes for the 3 deficiency allele phenotypes and the totals for both carriers and deficiency allele phenotypes with 95% CI in total populations of 19,300,000

	PISS	95% CI	PISZ	95% CI	PIZZ	95% CI	PIZ carrier and deficiency allele phenotypes	95% CI
AAT deficiency (non-COPD)	22,546	18,921–26,846	18,546	14,786–23,239	3,814	2,889–5,029	536,033	534,619–537,450
AAT deficiency (COPD)	27,649	21,445–35,586	71,563	56,410–90,634	46,306	37,097–57,708	1,829,673	1,827,152–1,832,198
	Odds ratio PISS	95%CI	Odds ratio PISZ	95%CI	Odds ratio PIZZ	95%CI	Odds ratio Total	95%CI
	1.227	1.205–1.248	3.869	3.807–3.932	12.168	11.733–12.576	3.666	3.655–3.678
Non-COPD AATD								536,033
COPD-AATD								1,829,673
Excess COPD-AATD								1,293,640

When the data for the PIMZ carrier phenotype is combined with the PISZ and PIZZ deficiency allele phenotypes to summarize the effect of the PIZ deficiency allele on the detection of AAT Deficiency among COPD and non-COPD patients, there is a significant increase, with an odds ratio of 3.666 in the number of AAT Deficiency subjects detected from 536,033 in the non-COPD cohort to 1,829,673 in the COPD cohort.

DISCUSSION

Overview

The statistical analysis of the combined non-COPD and COPD cohorts from genetic epidemiological studies performed by others in 6 countries (Australia, Canada, Germany, Italy, Sweden and the United States of America) have confirmed data already in the literature on the correlation between COPD and AAT Deficiency. The additional data provided in the present statistical analysis provides striking and new information on the relative changes in the normal and 2 deficiency allele frequencies as well as the numbers in the normal (PIMM), 3 carriers (PIMS and PIMZ), and 3 deficiency allele (PISS, PISZ, and PIZZ) phenotypic classes.

The numbers generated in the present study give a percentage of 2.8% (536,033/19.3 million) AAT Deficient subjects in an estimated population of 19.3 million non-COPD subjects, and a percentage of 9.5% (1,829,673/19.3 million) AAT Deficient subjects in an estimated population of 19.3 million COPD subjects if targeted screening for AAT Deficiency were performed in the United States of America. Thus, if, for example, screening tests for AAT Deficiency were performed on the estimated 19.3 million white COPD patients, a total of 1,293,640 additional AAT Deficient patients would be identified than such screening on a population of identical size of white non-COPD subjects. This also amounts to a 3.41-fold increase in the AAT Deficient patients detected.

Another estimate of interest is the 6.1% of AAT Deficient subjects of phenotype PISZ and PIZZ estimated in the population of 19.3 million white COPD patients (117,869/19.3 million)]*100). If we also include the number of AAT Deficient subjects with phenotype PIMZ, this percentage increases to 9.5% ([1,829,673/19.3 million]*100), whereas the number of AAT Deficient subjects of phenotype PIMZ alone is 8.9% ([1,711,804//19.3 million]*100).

Which Phenotypic Classes of AAT Deficiency Are at Risk for COPD in Earlier Studies in the Literature?

In an early study [[2]], a series of 111 index subjects with COPD who had forced expiratory volume in 1 second (FEV₁) of 70% or less of that predicted were matched on the basis of age, sex, occupation, and smoking history with control subjects who had an FEV₁ of 85% or more of that predicted. This study showed that PIMZ phenotypes in the index group outnumbered those in the controls by 8 to 5. In a follow-up study [[3]], PIMZ carriers of the PIZ deficiency allele accounted for 8.0 percent of the patients as compared to 2.9% of control subjects (p less than 0.0005), whereas the PIZZ phenotype was detected in 1.9% of the patients, compared to 0.04% of control studies performed by others.

In a more recent paper, the risk of AAT Deficiency and COPD for PIMZ was analysed by Hersh et al [[32]]. This group reported that case-control studies showed increased odds of COPD in PIMZ individuals, but this finding was not confirmed in cross sectional studies. Variability in study design and quality limits their interpretation. These results are consistent with a small increase in risk of COPD in all, or a larger risk in a subset of, PIMZ individuals. They concluded that future studies that evaluate the role of smoking are required to more conclusively determine the extent of the risk of COPD in PIMZ heterozygotes [[32]]. In a follow-up paper on 17 cross-sectional and case-control studies [[33]], the odds ratio for COPD in PIMS heterozygotes was 1.19 (95% CI: 1.02–1.38) and was not associated with COPD risk. However, the odds ratio for PISZ compound heterozygotes was 3.26 (95% CI 1.24–8.57), and was associated with COPD risk. Both findings are in agreement with the conclusions of the present study.

The Role of Smoking on Detection of AATD Among COPD Patients

We also intended to analyze the role of smoking on the detection of AAT Deficiency in both non-smoker and smoker patient cohorts as compared with non-smoker and smoker control cohorts. Unfortunately, in none of the over 50 papers selected from our PubMed literature searches dealing with COPD and AAT Deficiency were there data on individual cohorts of sufficient size for a meaningful Hardy–Weinberg Equilibrium statistical analysis. In addition, in none of these 50 papers were data presented in a manner that permitted the development of sets of cohorts that would have made it possible to evaluate the role of smoking on targeted screening for AAT Deficiency in white COPD and non-COPD populations.

Overall evaluation

Overall, the present study both confirms and extends the pioneering work on the correlation between COPD and AAT Deficiency published by Kueppers [[2], [34]] and his colleagues, and ten years later by Lieberman [[3]]. The larger sample sizes in both the non-COPD cohort and the COPD cohort in the present statistical analysis provided additional information on the relative risk of each of the 6 phenotypic classes of PIS and PIZ due primarily to the much larger sample sizes and the statistical power of these larger numbers.

CONCLUSIONS

The results of the present meta-analysis clearly support the targeted screening of white COPD patients for AAT Deficiency to enhance diagnostic recognition and treatment of AAT-deficient individuals. This approach also is supported by the conclusions of the Death Review Committee of the death of AAT Deficiency patients in the National Heart, Lung, and Blood Institute study [[20]], as well as, for example, another study to determine the efficacy of treatment of AAT Deficiency patients with Prolastin [[35]].

In addition, in the United States, those detected should take advantage of the patient resources for counselling and support groups developed by the Alpha-1 Association < http://www.alpha1.org/home/index.asp > for materials for patient education, as well as the patient AAT Deficiency registry, AAT Deficiency DNA tissue bank, developed by the Alpha-1 Foundation < http://www.alphaone.org/ >, and comparable facilities developed in Europe by the European Federation for Alpha-1 Antitrypsin Deficiency, < http://www.alfaeurope.org >.

The authors are indebted to Drs Marc Miravitlles, Pneumology Department, Hospital Clínic. Barcelona, Spain, and Maurizio Luisetti, Laboratorio di Biochimica e Genetica, Clinica di Malattie dell'Apparato Respiratorio, IRCCS Policlinico, San Matteo Pavia, Italy for their review and useful comments for the revision of the draft manuscript. The authors are also indebted to Dr. David Gelmont and Ms. Regina Ofiara, Baxter BioSciences, for their encouragement and financial support for the preparation of this report.