In this study, we report on the epidemiologic, demographic, clinical, pulmonary function, and serum biomarker data from the largest contemporaneous cohort of patients with autoimmune PAP assembled to date.
Several lines of evidence support the use of the term “autoimmune PAP” and the stratification of PAP into autoimmune and secondary forms. First, underscoring the importance of GM-CSF in pulmonary surfactant homeostasis, mice deficient in GM-CSF (6
) or its receptor (35
) develop a pulmonary phenotype biochemically, histologically, physiologically, and ultrastructurally identical to autoimmune PAP in humans (36
). Second, GM-CSF autoantibodies seem to be critical to the pathogenesis of autoimmune PAP because high levels are strongly associated with it but are not present in secondary or congenital PAP, other lung diseases, or in healthy individuals (17
). Their binding affinity for GM-CSF (~20 pmol/L) is higher than the GM-CSF receptor in its low- (~3,200 pmol/L) or high-affinity (120 pmol/L) binding state (37
), and they eliminate GM-CSF bioactivity in vivo
). Third, transfer of purified GM-CSF autoantibodies from patients with PAP into blood from healthy individuals reproduces the myeloid cell abnormalities observed in patients with autoimmune PAP (38
). Fourth, anti-murine GM-CSF antibodies reproduce these abnormalities in wild-type mice. Fifth, the courses of autoimmune and secondary PAP are different: Secondary PAP has a far worse outcome (H. Ishii, in preparation).
Important findings of this study are the incidence and prevalence data for autoimmune PAP in Japan. Using an intensive screening approach involving 98% of pulmonary physicians in one region, the highest incidence and prevalence estimates were 0.49 and 6.2 cases per million, respectively. Our prevalence value is higher than reported for Israel (3.7 per million) (39
) and lower than for the United States (~10–40 cases per million) (40
). One of 15 cases in the former study was congenital PAP, and the latter study included all three clinical forms. If we included all types of PAP identified in Japan during the study period, the prevalence would be 8.7 per million. We did not observe familial clustering in our study, in contrast to the Israeli report in which PAP occurred in two siblings. These likely represent congenital PAP, which occurs secondary to mutations in the genes encoding SP-B (41
), surfactant protein C (42
), or ABCA3 transporter protein (43
). Of the patients with autoimmune PAP in our study, 31% were asymptomatic and were identified by annual medical screening programs. Our observations suggest that the true incidence and prevalence of autoimmune PAP is higher than reported and show that nearly a third of cases are subclinical. Although regional differences in may exist, more studies are needed.
The demographics of Japanese patients with autoimmune PAP differ in several respects from a retrospective meta-analysis done by Seymour and colleagues, which includes most or all cases of PAP reported in the medical literature as of 2002 (2
). First, the median age at diagnosis was 51 years and similar in men and women, in contrast to Seymour and colleagues' report in which the median age was 39 years and was different in men and women (39 and 35 years respectively). Second, the age at diagnosis was normally distributed in women and did not have the bimodal pattern previously reported (2
). Third, the male:female ratio (2.10:1.0) and, fourth, the proportion of current smokers (28.5%) were lower than previously reported (2.65:1.0 and 72%, respectively) (2
). The absence of a male predominance among nonsmokers (never- and ex-smokers) in our study (male:female ratio = 0.60:1.0) is similar to the prior report (0.69:1.0) (2
) and is consistent with the possibility that the high proportion of men among patients with PAP may be explained by their higher frequency of tobacco use. However, a high proportion of women in our cohort had no history of smoking (83%) or occupational exposure (87%), suggesting that another factor may be involved in the etiology of autoimmune PAP. Our study did not address potential effects of passive smoke exposure.
It is surprising that COPD was not recognized more commonly in our cohort given that the proportion of current, ex-, and never-smokers was similar to the Japanese population in whom COPD occurs in 8.6% (44
). Using the same criterion (FEV1
/FVC < 0.70) as Fukuchi and colleagues (44
), only five individuals (2.7%) in our cohort had airflow limitation, whereas 24 individuals were expected of having airflow limitation. Of these, two were male (one ex-smoker, aged 55 yr; one never-smoker, aged 71 yr), and three were female (one current smoker, aged 39 yr; one ex-smoker, aged 28 yr; and one never-smoker, aged 44 yr). Although the reason for this is not clear, it is interesting that asthma, another common lung disorder with an inflammatory component of pathogenesis, was also underrepresented in our cohort (observed frequency = 2.4%; expected frequency ~8.2%). PAP may alter the phenotype of disorders with an inflammatory component of the pathogenesis. This is supported by observations that GM-CSF is required for myeloid cell functions in humans (38
) and mice (9
), where it regulates a number of innate immune responses, including the TLR4 response to lipopolysaccharide (46
), and GM-CSF autoantibodies in patients with PAP virtually eliminate GM-CSF bioactivity (18
). Thus, GM-CSF autoantibodies may blunt inflammatory responses in patients with PAP, which may affect tissue destruction in COPD and the tendency for exacerbations in asthma.
The DSS (22
) provided a useful measure of lung disease severity in symptomatic and asymptomatic autoimmune patients with PAP, which was important because nearly one-third of the patients were asymptomatic and because dyspnea is difficult to quantify in PAP due to the insidious onset. A limitation of this study was the absence of a dyspnea index. Notwithstanding, the DSS correlated well with the DlCO
% predicted, less well with FVC % predicted and VC % predicted, and not with FEV1
/FVC. Although PAP is usually described as a restrictive lung disease, reductions in lung volumes in autoimmune PAP were minor and fell in the normal range in most patients, suggesting that these pulmonary function measures may be of limited usefulness in assessing the severity of PAP lung disease. Physiologically insignificant restriction is further supported by the absence of hypoventilation, even in severe cases. Thus, in autoimmune PAP, hypoxemia is primarily due to reduced oxygen diffusion and possibly ventilation–perfusion mismatching.
Infections were less common among Japanese PAP registrants than previously reported (2
). Furthermore, although Nocardia
was identified in 60% of reported PAP cases complicated by infection (2
), no cases of Nocardia
infection were observed in our study during the period of observation. It is possible that these differences represent reporting bias or differences in clinical care of early infections because a number of the prior reports reflect infectious complications occurring over four decades.
Our observation that GM-CSF autoantibody levels did not correlate with disease severity as measured by the presence of symptoms, pulmonary function testing, or the DSS is consistent with prior reports (2
). Because GM-CSF autoantibodies in patients with PAP are polyclonal, it is possible that measuring the level of neutralizing antibody may provide a better correlation with disease severity. We have reported a patient with autoimmune PAP in whom serial measurement of serum GM-CSF neutralizing activity correlated well with disease severity (47
). Furthermore, the serum GM-CSF neutralizing activity was reduced in a patient who was successfully treated with inhaled GM-CSF (47
). Pulmonary compartmentalization of GM-CSF antibodies may be important in determining disease severity and could explain the lack of correlation with serum autoantibody levels. Neither the autoantibody levels, proportion of symptomatic individuals, pulmonary functions, nor DSS correlated with the duration of disease, which is consistent with the concept that disease severity does not worsen with time in most patients.
The method used to measure GM-CSF autoantibody levels is similar to prior reports (17
) except that a new monoclonal GM-CSF autoantibody standard was used. Although this standard yields reproducible results, it yields autoantibody levels one seventh that of the previous GM-CSF autoantibody standard isolated from pooled PAP serum (3
) and similar in sensitivity and specificity. Our results support the use of GM-CSF autoantibody measurement in the diagnosis of autoimmune PAP as an adjunct to chest CT and bronchoscopy.