Association between asbestos-related pleural plaques and resting hyperventilation.

This study reports an association between pleural plaques and resting hyperventilation in a group of workers exposed to asbestos. Information on exposure level, pack-years of cigarette smoking, chest radiographs, ventilation parameters, single-breath diffusing lung capacity, and arterial gases were obtained for 344 workers. After the exclusion of 37 workers for isolated parenchymal fibrosis, combined pleuroparenchymal fibrosis, or diffuse pleural thickening, 55 subjects with isolated pleural plaques were evaluated against 252 no-plaque workers. A quantitative pleural score revealed mild pleural disease. Forty-four workers with plaques (80%) had hypocapnia induced by resting hyperventilation. The quantitative pleural score correlated significantly with the partial pressure of carbon dioxide in arterial blood (correlation coefficient = 0.7). A decrement in forced vital capacity was associated with plaques, whether controlled for age, smoking, and exposure or not. It was concluded that the resting hyperventilation observed in some asbestos-exposed subjects is related to the presence of mild pleural plaques and a restrictive disorder.

amounted to 92% of all employees at the time of the study. Th e remaining 58 were retired workers, out of the 112 available from the retired work force. Attention was focu sed on 55 subjects (16 %) with isolated pleural plaques. Therefore, to eliminate potential confounders, we excluded 16 subj ects (4.7%) with isolated parenchymal radiological changes (profusion score ::;111 according to the cl ass ification of the International Labour Offi ce ), 17 subj ects (4.9%) with combined pleuroparenchymal disease, and four subj ects (1.1%) with diffuse pleural thickening from the anal ysis. Thus 55 workers with isolated pleural plaques were analyzed against 252 fellow workers without plaques. The participants gave their occupational and medical histories and underwent physical examinations, chest radiographs, spirometry, and other selected tests. The study was approved by the Ethical Committee of the Hospital, and informe d consent was obtained from all of the subjects.
Asbe sto s exposure was mainl y to chrysotile, but about 12% of the exposure was to croc idolite, also used dur ing the 1950s up to 1964. The factory products contained 84-86% cement and the rest was asbestos. Asbestos exposure data were avail able for the last 30 years . The exposures were assessed for each workplace on an annual ba sis. Du st particles were collected with an aspiration pump pro vided with filters of thin porous membranes of cellulose esters and counted by phase contrast micro scopy, as the number of fibers greater than 5 urn in len gth per milliliter.
The cumulative exposure to asbesto s in fiber-years for an indi vidu al wa s obtained as the concentrationweighted sum of exposures throughout the employment period, except for the 1950s. The exposure level adjusts D LCO to the normal Hb value of 14.6 g. dl' (12) and is a factor to correct for COHb induced by D Lco measurement (12). We did not measure COHb; instead we theoretically estimated an 0.7% COHb rise per each test performed (13), while when the FEV l o was <80% of the predicted value and the FEV% was <70%.
The single-breath diffusing lung capacity for carbon monoxide (D Lco) was measured with the subject seated (MASTERLAB®, Jaeger, Germany), and the predicted values were derived from Cotes & Hall (11). The average of the two acceptable tests was used. The measured D Lco values were consecutively corrected for carboxyhemoblobin (COHb), hemoglobin (Hb), and partial pressure of oxygen in alveolar air (PAoz) as follows: adjusts D l CO to the normal PA oz value of 105 mm Hg (13.97 kPa). This correction originates from our previous report [14 (equations 13 and 14)] by adaptation to traditional units and adoption of the normal value for the ratio of the pulmonary capillary blood volume (Vc) and the membrane diffusing capacity (Dm); that is, Vc/Dm was assumed to be 1.43 ml blood· min· mm Hg . ml' CO.
For all of the subjects arterial blood was sampled in duplicate via arterial lines placed in the radial artery. In order to avoid potential effects on ventilation from the insertion of arterial lines, samples were obtained about 20 min after insertion. Xylocain was where for the 1950s was taken to be 3 fibers· ml'. Roughly, the average concentrations were about 3 fibersml-1 in the 1960s, 2 fibers· ml' in the 1970s, and 1 fiber· ml' in the 1980s. The factory was closed in 1989. Smoking and respiratory histories were taken as proposed by Ferris (7). A standardized questionnaire on respiratory symptoms was carried out by a trained chest physician. Dyspnea was evaluated only as a dichotomous variable, since a morereliable related parameter, partial pressure of carbon dioxide in arterial blood (Pa coz)' was being measured. The anthropometric characteristics, smoking habits, and exposure level of the workers are shown in table 1 according to the presence or absence of pleural plaques. The mean fiber-years was 32.7 (range 2-78), and 41.9% were current smokers with mean pack-years of 31.4 (range 3-65). The mean pleural score was 13.6 (range 4-51). The group with plaques was significantly older and more heavily exposed, included a larger proportion of smokers, and had a greater mean of pack-years of cigarette smoking.
Standard posteroanterior radiographs at full inspiration were taken and read independently by two radiologists trained in the system of the International Labour Office (ILO) for classifying radiographs (8), without knowledge of the epidemiologic, clinical, and lung function data. Their average ILO score was adopted. Pleural plaques were defined when a plaque-like thickening at the lung pleura interface along the lateral thorax or either hemidiaphragm was 2 mm or more. Quantitative pleural scores were obtained according to Oliver et al (2).
Forced vital capacity (FVC), forced expiratory volume in I s (FEV l o), the FEV% [(100· FEVlo)1 FVC], and the maximal expiratory flow rates at 25 and 75% of the FVC (FEF z5_75 % ) were measured with a spirometer (Mijnhardt VICATEST®, The Netherlands) using standard criteria of the American Thoracic Society (9). The largest value of three acceptable maneuvers was used. Predicted values were obtained from Cotes (10). Minute ventilation (V E ) , breathing frequency (f), and tidal volume (VT) were also obtained on the same equipment. Restriction was defined as an FVC of < 80% of the predicted value and an FEV% of~70%. Obstruction was assumed used as the local anesthetic; the volumes of drawn blood ranged from 1.5-2 m!' The average values of the two measurements of Pa cor PaOl' and pH were adopted (ABL-I, Radiometer, Denmark ). The machine used for the measurements had a built-in automatic calibrating set for periodic checks every 20 min. Hypocapnia was assumed when the value of Pa COl was less than 35 mm Hg (4.66 kPa), while the normocapnic range was defined as a Pa COl of 35-45 mm Hg (4.66-5.99 kPa). The subjects with plaque s were not aware of their plaque status.
We emphasize that the no-plaque group should not be considered strictl y as a reference group in the statistical sense, but rather as a comparative group of fellow workers from the same factory. Since the two group s were formed by the discriminative criterion of having pleural plaque s proved to be different regarding many anthropometri c and exposure parameters, besides simple univariate tests, the remaining associations between the presence of plaques and pulmonary function were examined by multivariate tests, potential confounders being controlled for.
The test results were expre ssed as means and standard deviations. For the univariate analyses the Mann-Whitney and chi-square tests were used to evaluate the significance of the differences in the pulm onary function tests, arterial blood gases, and prevalences of restriction and obstruction between the plaque and no-plaque groups. To evaluate the associations between pleural plaques and the various functional parameters that remained after control for the potential confound ers of age, asbestos exposure, and smoking, a logistic regression was used. The relation between the quantitati ve pleural score (QPS) and the Pa C02 was assessed by linear regression. A P-value of <0.05 was considered significant.

Results
For the workers with plaques the mean FVC was 75.8 % of the predicted value, the mean FEF 25_75 % was 88.2% of the predicted, the mean FEV% was 114.7% of the predicted, and the mean total lung capacity was 89.3% of the predicted. All of these values were significantly different from those of the workers without plaques (P x l fr", P = O.OOOl , P<lO-6, and P = 0.00000 2, respecti vely) (table 2). Both the mean measured D LCO and corrected D Lc o were significantly lower in the group with plaques than in the no-plaque group (P < 10-6 and P =0.003, respectively) (table 2). Forty-two percent in the plaque group (23 of 55 workers) had restrictive disorder, compared with 16.7% of the no-plaque group (4 1 of 252 workers) (P = 0.00005). Thus the risk ratio for having a pulmon ary restriction between the plaque and noplaque group can be assessed as 23:55/4 1:252 =2.57.
As for the obstruction, the two groups did not differ significantly (P = 1.0) (table 2).
For the workers with plaque s the mean breathing frequency was 20.2 breaths· min-I, a value which is 1.41 times greater than the 14.3 breaths-mirr ' of the no-plaque group (P < 10-6 ). Tidal volume (VT) was also larger in the plaque group (P = 0.005). Consequently tidal volume, being the produ ct of f · VT' was significantly larger in the group with plaques (P< 10-6). As a result of resting hyperventilation , Pam and pH were increased, while Pa C02 was decreased in the plaque group when compared with the noplaque group (P < 10-6, for each comparison), and hypocapnia was present more frequently in the subjects with plaques (X 2 =232, P< 10-6) (table 3). Dyspnea was reported by 74% of the workers with plaques (4 1 of 55 workers), which was a much higher prevalence than in the no-plaque group (12% or 30 of 252 workers) (X2 =96.5, P< 10-6).
In the logistic regression analysis, after control for age, pack-years of cigarette smoking, and fiber-years of asbestos exposure, the presence of hypocapnia (P< IO~) , decrements in FVC (% predicted) (P = 0.03), and total lung capacity (% predicted) (P = 0.08), as well as the increment in the FEV% (% predicted) (P =0.04), remained significantly associated with the presence of pleural plaques (table 4). Dec- • rements in the measured D LCO (% of age, height, and gender predicted) and increments in the breathing frequency revealed a significant association with the presence of plaques, after control for smoking and exposure (P = 0.04 and P = 0.00002, respectively).
However, DLcocorrec,ed (% of predicted) was not found to be significantly associated with plaques (P = 0.07) (table 4). The logarithm of QPS exhibited a stronger linear relation to Pa C02 than did the QPS itself (correlation coefficient -0.70 versus -0.62) (figure 1).

Discussion
Exposure to asbestos can lead to a restrictive ventilatory disorder, a decrease in lung static compliance, and a reduction in lung diffusing capacity for carbon monoxide (15,16). Recent longitudinal studies have reported transient increases in D LCO and FEF Z5 _ 75 % as the earliest functional abnormalities, which are followed by normalization and ultimate reduction in some asbestos-exposed subjects (17,18). This dynamics of pulmonary function indices may account for the large range of the values that are usually observed in cross-sectional studies, depending on the prevalence of the subjects in various stages of asbestosis. The few cross-sectional studies available had shown that some functional abnormalities can be present in subjects with normal chest radiographs (19,20).
In the present study we observed a resting hyperventilation in 80% of the asbestos-exposed workers with radiographic evidence of pleural plaques. None of the control variables of age, smoking, or exposure entered significantly into the assessment of an association of plaques with resting hyperventilation. Previously, resting hyperventilation was found to be associated with asbestos exposure only in cases with progressive parenchymal and pleural fibrosis leading to arterial hypoxemia and increased chemical respiratory drive (5). A similar response was also reported for patients with chronic interstitial lung dis-ease (6,21). Since our plaque group was normo xemic , chemical humoral factors can be ruled out as the cau se; thu s the pos sible mechanisms of resting hyperventilation are mechanical adaptation to the increased ela stic work of breathing or a reflex-mediated increase in breathing frequency originating from the che st wall receptors (22). Quantitative support for the hypothesis that pleural plaques provoke resting hyperventilation is the observed significant linear correlation between the pleural score and Pa C02 (correla tion coefficient -0.70).
The analysi s of the functional imp airment observed in subj ects with radiological pleural changes is becoming an important issue since a large number of workers with a low-to-moderate expo sure history have asbestos-associated pleural disease (23,24). The restrictive ventilatory disorder observ ed in 42 % of our subjects with plaques was characterized by a reduction in FVC and an increase in the FEV%, since FEVl.o was in the normal range. The variable smoking did not enter significantly into the logistic regression of plaque entity versus restriction entity, and this finding is physiologically plausible. Generally, smoking can be related to obstruction, but such a relation was not observed in this study . However, the variable exp osur e entered significantly into the multi variate analy sis of plaqu e-restriction associ ation. Therefore, a plausible conclusion can be drawn that pleural plaques and the associated restrictive disease were due to cumulative asbestos exposure. The mean D LCO (% of predicted), whether corrected for COHb and adjusted to normal Hb and Paozor not , in the plaque group was significantly lower than in the no-plaque group. However, after control for smoking and exposure, D LCOcmrecte<! (% of predicted) prov ed to have a less than significant association with plaques. Thi s finding suggests that plaque s and D LCO deterioration are relati vely independent entities, but it also indicates the possibility that, in the work ers with plaques, the D LCO decrease could have been an early sign of coexisting interstitial lung disease. However, mild D LCO reduction in our subjects with plaques was not responsible for the resting hyperventilation, as was evident from their increased Pa 0 2 ' which pro ved that the aiveocapillary membrane was not an obstaele to gas exchange. Measuring the D LCO components membran e diffu sing capacity and capillary blood volume might pro ve useful in gaining some insight into the mechanisms involved. Oliver et al (2,3) reported restricti ve functional impairments in railroad workers, a finding which is in accord with our results, exc ept for D , which was less affected in their sample. In LCO • addition our study included arterial gas determinations and revealed a strikingly high prevalence of hypocapnia in the asbe sto s-exposed work ers with plaques.
As for the potential shortcomings of the study, there is a lack of computed tomography results to help exclude earl y parench ymal di sea se. As already mentioned, this lack should not invalidate the as-sessed associati on between plaques and hyp erventilation, but its correction might lead to some insight into the associ ation between plaqu es and restri ct ion . Another probl em is that expo sure in the 1950s was only estimated and not measured and that some crocidolite was used until 1964. If, in consequence, the exposure of the older plaque workers was und erestimated, the estimate of the ass ociation betw een plaque s and res ting hyperventilation would not be affected, but the es timate of the plaque-restriction ass ociation would .
In conclusio n, resting hyperv en tilation was observed in a significa nt portion of our asbe sto s-exposed subje cts and wa s related to pleural plaques. The pleural plaques them sel ves appeared to be related to restricti ve impairments.