This study was designed as a case-control study, and was approved by the Ethics Committee of the State University of Rio de Janeiro. Informed consent was obtained from all volunteers before inclusion in the study. Healthy control subjects with normal spirometry who had never smoked, as well as smoking subjects who were on no regular medications and had no allergic, respiratory, cardiovascular, gastrointestinal, renal or neurological symptoms were recruited among people working in the laboratory and from the State University of Rio de Janeiro area. All subjects had stable health for at least four consecutive weeks and signed written informed consent. Baseline data, including age, height and weight, were obtained from each subject at the time of the procedures.

The amount of tobacco smoked and the duration of smoking were quantified using the number of pack-years 20. Verbanck et al. 21, using the multiple breath washout test, showed that the earliest signs of small airway malfunction in smoking patients are seen from 10 pack-years onwards. Therefore, we studied smokers with tobacco consumptions near 10 pack-years who were recruited from both university personnel who smoke and from patients who visited the smoking cessation clinic of our university hospital. All smokers were current smokers. These volunteers had been instructed to abstain from smoking for at least two hours before the testing in order to avoid acute effects of tobacco use on lung function 22.

Total respiratory resistance and reactance were measured using a forced oscillation system, which has previously been described in detail 2324. The system was a standard multifrequency FOT test system operating in the most commonly used frequency range 710. Measurements were conducted in conformity with the recommendations issued by a task force of the European Respiratory Society 10. Briefly, small amplitude pressure variations from 4-32 Hz generated by a loudspeaker were applied at the mouth using a mouthpiece. The pressure input was measured with a Honeywell 176 PC pressure transducer (Honeywell Microswitch, Boston, MA, USA), and the airflow was measured with a screen pneumotachograph connected to a similar pressure transducer with a matched frequency response. These signals were digitized at a rate of 1024 Hz for periods of 16 s by a personal computer. Usually a sampling frequency of 128 Hz is used when the maximum frequency is 32 Hz. In the present study, a higher sampling frequency was used in order to provide a detailed visual description of the artefactual events, such as swallowing or glottis closures, on the flow and pressure signals. As recommended 10, these signals are displayed on the screen during the measurement. The fast Fourier transforms (FFT) of the digitized signals were computed using blocks of 4096 points with 50% overlap. Overlapping was used in order to increase the number of independent data blocks and reduce random errors. Three measurements of 16 s each were made, and the result of the test was calculated as the mean of these measurements.

Resistive impedance data were subjected to linear regression analysis over the frequency range from 4 to 16 Hz. Previous studies from our group 1825 showed that most of the changes due to airway obstruction are observed over this frequency range. The resistive impedance at 0 Hz (R0) was extrapolated from this analysis. This parameter is related to the total resistance of the respiratory system and is usually used as an index of airway obstruction 26. The mean resistance (Rm), sensitive primarily to airway caliber 8, was also calculated for the frequency range from 4 to 16 Hz. The slope of the resistive component of the respiratory impedance (S) (which is associated with respiratory system homogeneity 1727) was also obtained from the analysis. Negative values of this parameter reflect abnormal patterns of ventilation distribution. Previous works reported S values near zero in healthy subjects 232428.

The properties of energy accumulation were characterized by the mean reactance (Xm), a property usually related to respiratory system non-homogeneity 29. This parameter was calculated based on the entire studied frequency range (4 to 32 Hz). These properties were also characterized by both the resonance frequency (fr), which is defined as the frequency at which the Xrs equals zero, and the respiratory system dynamic compliance (Crs, dyn), which was estimated using the Xrs at 4 Hz (Crs, dyn = -1/(2πfXrs)) 30. The same frequency was used to evaluate the absolute value of respiratory impedance (Z4Hz). This variable represents the total mechanical load of the respiratory system 78910, and is associated with the necessary work to promote the movement of air in the respiratory system. In the present work, Z values measured at 4 Hz mainly describes the resulting effect of the total resistance and respiratory system compliance.

To perform the FOT analysis, the volunteer remained in a sitting position, keeping his or her head in a normal position and breathing at functional residual capacity through a mouthpiece. During the measurements, the subjects wore a nose clip and firmly supported their cheeks and sub-mandible tissue with their hands 101825 (Figure 1).

The validity of the data was measured by computing the coherence function, which evaluates the signal-to-noise ratio. This function is the equivalent in the frequency domain of the correlation coefficient used in the time domain. In order to minimize the effect of noise associated with the breathing signal, only values with a coherence function of 0.9 or more were considered adequate 162631. Any time the computed coherence was less than this threshold, the maneuver was considered invalid and the exam was repeated. Whenever adequate coherence measurements could not be obtained according to these criteria, the patient was excluded from the study 1825. The data acquisition started 30 seconds after the beginning of the exam in order to allow the volunteer to be accommodated to the equipment, and to guarantee the regularity of the respiratory pattern during the exam 10.

Measurements of Forced Vital Capacity (FVC), Forced Expiratory Volume for the first second (FEV₁), FEV₁/FVC, and the ratio of Forced Expiratory Flow (FEF) between 25% and 75% of FVC to FVC (FEF/FVC) were obtained for patients in a sitting position using a closed circuit spirometer (Vitrace VT-139; Pro-médico, Rio de Janeiro, Brazil). These parameters were presented as raw data and percentiles of the predicted values (%). Predicted values for spirometry were obtained from Knudson et al. 32 and Pereira et al. 33. Forced expiratory maneuvers were repeated until three sequential measurements were obtained. The indices studied were those obtained through the better curve, which was selected based on the higher value of FEV₁ plus FVC. Quality control of spirometry is given by the ATS criteria, with the software allowing the detection of non-acceptable maneuvers.

The volunteers were stratified, and when the achieved data presented a statistically normal distribution, smokers and controls were compared using a Student's t-test. A non-parametric test (Mann-Whitney U Test) was applied when the data did not present a normal distribution. These analyses were performed using STATISTICA 5.0 software. The results are presented as mean +/- standard deviation. A p value of less than 0.05 was considered statistically significant.

The number of pack-years was calculated by multiplying the mean number of packs (one pack = 20 cigarettes) consumed daily by the number of years that the subject had their smoking habit 20. These data were collected via a questionnaire, which was interviewer-administered. Although questions on tobacco consumption may not precisely quantify respiratory changes, it is widely known that smoking introduces inflammation and narrowing of peripheral airways. Therefore, the presence the smoking habit was used as a reference in this case-control study to separate the two studied groups. The performance of the FOT indices in the detection of smoking-induced respiratory alterations in the studied sample was evaluated by means of receiver operating characteristic (ROC) curves 3435, This method is able to identify the optimal cut-off point that discriminated most efficiently between the absence and presence of early respiratory changes in the studied sample. Analysis of ROC curves is performed by plotting sensitivity versus 1-specificity for each possible cut-off level. This way, the larger the area under the curve (AUC), the more valid the diagnostic test is. According to the literature, ROC curves with AUCs between 0.50 and 0.70 indicate low diagnostic accuracy, AUCs between 0.70 and 0.90 indicate moderate accuracy, and AUCs between 0.90 and 1.00 indicate high accuracy 3436. Goedhart et al. 37 considered 0.7 to be a good cut-off value for a useful discriminator for clinical use. In the present study, we considered 0.75 to be the minimum value of the AUC for adequate diagnostic accuracy. The ROC curves were constructed using MedCalc 8.2 (Medicalc Software, Mariakerke, Belgium).

The comparison of the performance of FOT and spirometry to identify early respiratory alterations in the studied sample was conducted considering the AUC obtained from these two techniques, following the theory described by Metz 38. MedCalc 8.2 was used in these comparisons. The values of sensitivity, specificity, and AUC were obtained based on the optimal cut-off point, which was chosen in order to balance the values of sensitivity and specificity 34.

The required sample size was calculated in order to allow for the comparison of a clinically useful area (AUC ≥ 0.75) with a null hypothesis value (AUC = 0.5, meaning no discriminating power). This was calculated using the software MedCalc version 8.2, and assuming Type I and Type II errors of 20% 39. The minimal required sample size resulted in 24 volunteers in each group.

The two studied groups were of comparable age, weight and height, showing only non-significant statistical differences (Table 1). The study conducted by Verbank et al. 21 showed early changes in small airways in smokers with normal spirometry and as little as 10 pack-years of smoking history. The mean tobacco consumption of the studied smokers here was 11.2 ± 7.3 pack-years, characterizing a sample of smokers with early respiratory changes. In line with this hypothesis, we observed only small and non-significant reductions in the spirometric parameters of the smoking group (Table 1).

In line with previous results 5151617, we found that higher resistances are observed among smokers (Figure 2). A uniform increase was observed when comparing Rrs among the control group and smokers. This increase is consistent with that recently reported by Ionescu et al. 43 in a model study describing the effects of COPD on respiratory resistance. In close agreement with previously published results 1516, we found that smokers displayed more negative Xrs values than healthy subjects (Figure 2B). Smoking provokes in its initial stages adverse reaction from the airway cells to the noxious components, which results in increases of small airways resistance. This is associated with a decrease in the dynamic compliance 44, which, in its turn, results in more negative values of Xrs. These results are associated with an increase in respiratory system non-homogeneity and in close agreement with the model study conducted by Ionescu et al. 43.

The significant increase in R0 (Table 2, p < 0.001) may be associated with airway obstruction 26. Since it is known that the effects of smoking begin in the peripheral airways, we can speculate that this increase is associated with this section of the bronchial tree. The increase in R0 values for smoking subjects was consistent with previously published results 515161743 and suggests that R0 values could be useful to detect initial airway obstructions associated with smoking. In order to investigate this possibility, ROC curves were made (Figure 3). Although a variety of summary indices have been used to measure the performance of diagnostic systems 4546, the area under the ROC curve (AUC) has the clinically useful interpretation of representing the probability of correctly discriminating between two subjects in a randomly selected pair of abnormal and normal subjects 3947. In the present study, 0.75 is considered the cut-off value for a useful discriminator. Thus, R0 reached acceptable values, identifying the respiratory changes due to a tobacco consumption near 10 pack-years with a sensitivity of 71.4% and a specificity of 75% (Table 3).

It is important to point out that it is possible that some people smoke for 10 years and do not actually have any structural change going on. Therefore, the results described in Figure 3 are limited to the studied sample. In spite of this limitation, these results confirmed that FOT can be used to detect the early alterations in the respiratory function presented in this sample due to the smoking process. This suggests that this technique could easily and non-invasively detect early airway obstruction in these patients.

In agreement with the present work, Hayes and colleagues 15 found no significant difference in Rm comparing non-smokers and smokers. This suggests that Rm would not be useful in detecting early changes due to smoking. Confirming this hypothesis, this parameter did not reach an AUC value that was considered useful for clinical use by the criteria used in this study (Figure 3, Table 3).

Although we can observe a clear difference in the S parameter between the controls and the smokers, this difference was not statistically significant (Table 2). It probably happens due to the high SD presented by this parameter. In agreement with these results, we observed a small AUC (Table 3), suggesting that this parameter may not be sufficiently sensitive to detect the early changes in the respiratory system of smoking patients.

Tobacco use was associated with a significant increase in fr (Table 2; p < 0.002). This links tobacco consumption with decreased respiratory system homogeneity and suggests that fr is useful as an index for the initial effects of smoking in the studied sample. The results described in Figure 3 and Table 3 shows that, although this parameter reached a borderline value of AUC, it may not be considered useful for the identification of the initial respiratory effects of smoking.

The low pack-year value of the smoking group resulted in decreased Xm mean values (Table 2, p < 0.0009). This reflects the impact of smoking on the reduction of respiratory system homogeneity and dynamic compliance of the studied subjects. The results described in Figure 3 and Table 3 show that, similar to fr, this parameter reached a borderline value of AUC, and may not be considered adequate.

Tobacco smoking resulted in a decline in Crs, dyn (Table 2, p < 0.001), which could be associated with an increase in peripheral airway resistance or with a reduction in the respiratory system compliance. Our results are in agreement with the findings of Verbank et al. 21 obtained in smokers with normal spirometry and as little as 10 pack-years of smoking history that indicated early changes of small airways in both the conductive and acinar lung zone compartments. In the studied sample, this parameter reached acceptable values for clinical use, identifying respiratory modifications with a sensitivity of 78.6% and a specificity of 67.9% (Figure 3 and Table 3).

As can be seen in Table 2, the modifications in resistive and elastic properties of smoking patients resulted in significant increases in Z4Hz. This parameter is related to the total mechanical load of the respiratory system. Therefore, it may be associated with fatigue and breathlessness in smoking patients. In agreement with the promising results of the initial analysis described in Table 2, we observed that this parameter achieved an adequate AUC to detect early respiratory alterations in the studied sample (0.79). In fact, Z4Hz was the most adequate measure to use to correctly identify the initial respiratory modifications in the studied smoking patients, with a sensitivity of 75% and a specificity of 75% (Table 3).

Recent recommendations for research in COPD 42 include the need for improved noninvasive mechanical tests of lung function. The present study was conducted as an effort to contribute in this direction, and showed that R0, Crs, dyn and Z4Hz are very promising parameters to non-invasively evaluate early modifications due to smoking. In the studied sample, these parameters were adequate to detect respiratory alterations even in conditions of small changes in spirometry. This suggests that the FOT test may be an useful screening tool in the management of smoking-induced lung disease, offering the possibility of showing abnormalities at a time when pathologic changes are still potentially reversible 21.

According to Metz 38, the AUC is usually the best discriminator when we have a number of ROC curves to compare. In the studied groups, these analyses were clearly in favor of FOT parameters (Table 4). The results show that the AUC of R0 was significantly larger than that obtained for FEV₁(L) and FEV₁(%). These parameters are standard measures of lung function commonly used in the evaluation of patients with COPD 19. The performance of R0 was also significantly higher than that of the FEF_25-75(L) and FEF_25-75(%), and borderline differences were achieved considering FEV₁/FVC (%) and FEF/FVC (%).

There were no statistically significant differences in the AUC for the S parameter and the spirometric parameters, probably due to the high SD associated with these measurements. The diagnostic accuracy of fr and Crs, dyn was higher than all of the spirometric parameters except FEV₁/FVC. Note that borderline differences were also achieved in these comparisons. On the other hand, the diagnostic performance of Z4Hz was higher than all of the spirometric parameters. These promising results suggest that R0, fr, Crs, dyn and Z4Hz values could be useful parameters in detecting early changes associated with smoking.

These results are in close agreement with previous studies that included asthmatic patients with normal spirometry, in which FOT parameters were useful to identify the initial respiratory modifications in these patients 25. In more recent work, conducted in patients with sarcoidosis, it was observed that FOT parameters were adequate to detect respiratory alterations, even in conditions of normal spirometry 48.

There are two potential limitations in the present study. First, we tested the hypothesis that, in a group of otherwise healthy current smokers, the forced oscillation technique was useful to detect the early effects of smoking on respiratory mechanics. The study was conducted comparing non-smoking and smoking subjects with low pack-years. A questionnaire was used as a reference in order to separate smokers and non-smokers. One could argue that a gold-standard technique was not used in this work to define the early changes in respiratory mechanics. However, as pointed out by Verbank et al. 21, in the early stages of smoking-induced lung disease, a gold-standard of peripheral lung damage is as yet impossible to obtain in human subjects. As it is widely known that small respiratory abnormalities are typical in subjects with low exposure to cigarette smoke 19, we compared the performance of FOT and spirometry in detecting these abnormalities in the same group of patients. It is important to point out that the respiratory changes observed in the present work are coherent with that obtained in a morphological model 43, confirming the consistency of these results.

The second limitation is that smoking history is difficult information to collect due to recall bias and changes in the amount of smoking throughout a lifetime. Thus it is possible that the value of pack-years in the smoking group is not very accurate. Since caution was taken to make FOT and spirometric measurements in the same subjects, this uncertainty equally affects the two measures. Therefore, this uncertainty is not a problem in the present study.