Abstract
This study was designed to assess the effect of differential leukocyte depletion during chemotherapy by monitoring the levels of exhaled hydrogen peroxide H2O2 and nitric oxide (FeNO) present.
In 39 patients with lung cancer (chronic obstructive pulmonary disorder up to stage II, median forced expiratory volume in one second 78% predicted), measurements were performed before a cycle of therapy (day 1), at least once during the cycle (day 8: n = 34; day 15: n = 19), and afterwards (days 21–29).
There were significant changes in the level of H2O2, FeNO and peripheral blood cell differentials over the visits. The level of H2O2 was decreased only on day 15, with a median (difference between the upper and lower quartiles) fall of 31 (57)%, while FeNO was reduced only on day 8, by 22 (40)%. Neutrophil numbers were unchanged on day 8 and decreased by 59 (48)% on day 15, while monocyte numbers were decreased on day 8 by 87 (39)%. On days 21–29, values had returned to baseline.
Taken together with previous findings, the parallel course of levels of exhaled hydrogen peroxide and neutrophil counts suggests that a major part of exhaled hydrogen peroxide is due to neutrophils via the conducting airways. In contrast, the production of exhaled nitric oxide seems to be primarily associated with monocytes.
Hydrogen peroxide (H2O2) in exhaled breath condensate (EBC) has been used over the last 15 yrs as a marker of airway inflammation 1. Elevated levels of H2O2 were found in chronic obstructive pulmonary disease (COPD) 2–4, bronchiectasis 5, asthma 6–9, pneumonia 10 and cystic fibrosis 11, and these levels were responsive to anti-inflammatory 12–14, antibiotic 11 or antioxidative interventions 15, 16. The flow dependence of H2O2 levels suggests that the conducting airways are the major site generating the exhaled H2O2 17. In addition, there are arguments that an important cellular source of exhaled H2O2 might be neutrophils, as suggested by the correlation with disease severity in COPD 2 or the findings in bronchiectasis 5. However, inflammatory airway diseases are known to be associated with activation of various cell types. This introduces some difficulty in attributing the exhaled H2O2 to a specific cell type by correlation of baseline values in diseases of different severity. Similarly, antioxidative or anti-inflammatory interventions are likely to exert rather broad effects. In healthy subjects, more specific evidence on the cellular source of H2O2 has been provided by the correlation between baseline levels of H2O2 and the ability of blood neutrophils 18 or phagocytes 19 to become activated.
Based on this, ways to elicit effects on various cell types in a more selective or differential manner were explored. Furthermore, patients with no more than mild-to-moderate inflammatory airway disorders would probably show a more stable pattern of cellular activity over time than patients with severe disease, and a trial using an intervention and longitudinal analysis would keep patients' characteristics comparable between measurements. A major effect on leukocyte numbers is known to be elicited by chemotherapy, as used in the treatment of lung cancer. Various cell types are differentially affected, but the effect on neutrophils is by far the predominant one and can even reach the clinically significant state of neutropenia.
The present authors therefore measured exhaled H2O2 in patients undergoing chemotherapy for lung cancer to assess whether its level would change in parallel with neutrophil numbers in peripheral blood. Measurements were performed before, at least once during, and after a cycle of chemotherapy. For comparison, fractional exhaled nitric oxide (FeNO) was also determined.
SUBJECTS AND METHODS
Patients
Thirty-nine participants without asthma and with COPD of, at most, stage II 20 were recruited from consecutive patients undergoing chemotherapy for lung cancer (table 1⇓). Ten patients were Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage 0, eight patients stage I, and 21 patients stage II (median (interquartile range)) forced expiratory volume in one second (FEV1) of this group: 72.3 (5.3)% predicted (% pred); lowest value 59 % pred). Three patients were atopic but did not have relevant allergen exposure. Five patients had small cell lung cancer, 33 had nonsmall cell lung cancer and one had mesothelioma. Tumour stages were Ia (n = 2), Ib (n = 1), IIIa (n = 2), IIIb (n = 11), IV (n = 21) and other (n = 2). The study was approved by the local ethics committee and patients gave their written informed consent.
Patients' characteristics assessed upon inclusion (day 1)
Study design
Baseline data were obtained within 2 h before a cycle of chemotherapy (day 1). In 16 patients, this was the first cycle, in six the second, and in 17 a later cycle. Depending on their individual time schedule, patients visited the laboratory again on day 8 and/or day 15. To check whether values had returned to baseline, patients also visited the laboratory between day 21 and 29 (median 25) after chemotherapy. All measurements were performed in the morning within a time interval of 2 h to circumvent circadian variations.
Chemotherapy was based on carboplatin, cisplatin or other drugs in 24, 12 and three patients, respectively, and chosen independently from this study. Dexamethasone was given as antiemetic medication (12 mg·day−1) over 1, 3 or 5 days in one, two or 12 patients, respectively. Six patients received lenograstim (glycosylated granulocyte colony-stimulating factor, Granocyte®; Chugai Pharma Europe Ltd, London, UK; dose 33.6 IU) after day 15 to counteract the fall in white blood cell numbers.
Methods
EBC was collected during 10 min of tidal breathing (ECoScreen; Viasys, Höchberg, Germany) while patients were wearing a nose-clip. The subjects were asked to remove the mouthpiece frequently and to swallow in order to avoid saliva accumulation in the mouth. After collection, the EBC was quickly thawed. p-Hydroxyphenylacetic acid (PHPA) and peroxidase were immediately added to a 300-µL aliquot of EBC 17, which was mixed and then stored at −20°C. The reaction between H2O2 and PHPA forms a stable fluorescent product that can be stored for at least 1 yr. Standard solutions with known H2O2 concentrations were treated in the same way and were used to check the fluorometric measurements. The limit of detection was 0.03 µM. Blood differential cell counts (SF-3000; Sysmex, Hamburg, Germany) and FEV1 21 were assessed by standard methods. FeNO was measured at a flow rate of 50 mL·s−1 22 using a chemiluminescence analyser (NIOX; Aerocrine, Solna, Sweden). Ambient NO concentrations never exceeded 10 ppb.
Data analysis
Data are presented as median values (interquartile range). Comparisons of visits were performed using Friedmans' nonparametric ANOVA, pairwise comparisons by the Wilcoxon signed-rank test, and comparisons of groups by the Mann–Whitney U-test.
All patients (n = 39) were measured before the start of chemotherapy (day 1) as well as afterwards (days 21–29). As only 14 patients could attend both visits at day 8 and day 15, two separate analyses were performed, comparing either day 8 with day 1 and days 21–29 (n = 34), or day 15 with day 1 and days 21–29 (n = 19) to keep the loss in statistical power small. As the groups having attended the laboratory on day 8 versus day 15 were comparable (table 2⇓), and similar changes were found in the patients who had attended all visits, it was considered justified to pool all data for the purpose of illustration (fig. 1⇓).
Time course of the concentration of hydrogen peroxide (H2O2) in a) exhaled breath condensate (EBC) and b) the fraction of exhaled nitric oxide (FeNO), and the number of c) neutrophils, d) monocytes and e) lymphocytes in peripheral blood. The horizontal axis represents the four consecutive visits. Median values and interquartile ranges are shown. For eosinophils, see table 2⇓. To increase the reliability of statistical parameters, the graphs comprise all available data and therefore represent different groups of patients at day 1 and days 21–29 (n = 39) versus day 8 (n = 34) versus day 15 (n = 19). As the values in these groups were very similar (see table 2⇓), we compiled the data in this way for the purpose of data visualisation. f) illustrates that the pooling was not misleading. It depicts median values of H2O2 (a), FeNO (b), neutrophils (c) and monocytes (d) in those 14 patients who attended all four visits; scales are the same as in a–e.
Results from laboratory visits of patients
Variability was quantified by intraclass correlation coefficients via ANOVA. To facilitate a comparison with therapy-induced changes, it was additionally expressed as the absolute difference between the values observed at day 1 and days 21–29 relative to the mean of these values. Correlation analysis was performed by Spearman rank correlation. Statistical significance was assumed for p<0.05.
RESULTS
Changes in blood parameters
When comparing days 1, 8 and 21–29, total leukocyte and neutrophil numbers showed no significant changes. In contrast, there was a difference in monocyte numbers (ANOVA, p<0.0001), which were lowered at day 8 (table 2⇑). The percentage of monocytes was reduced at day 8, whereas that of neutrophils was raised (p<0.001 each).
When days 1, 15 and 21–29 were compared, there was a fall in leukocyte, neutrophil and monocyte numbers at day 15 (ANOVA, p<0.05 each; table 2⇑). The percentages of monocytes and neutrophils did not differ significantly between these three visits. Similarly to monocytes, changes in eosinophil and lymphocyte numbers occurred on both day 8 and day 15 (ANOVA, p<0.05 each); the same was true for the percentage of eosinophils, which was lowered.
The median value (interquartile range) of the fall in leukocyte number on day 15 relative to day 1 was 2.38 (3.04) ×1,000·µL−1 or 41 (30)%. The neutrophil number was reduced by 1.78 (2.72) ×1,000·µL−1 or 59 (48)%. Monocyte numbers were reduced by 0.42 (0.31) ×1,000·µL−1 or 78 (39)% at day 8 and by 0.13 (0.45) ×1,000·µL−1 or 26 (51)% at day 15, whereby changes were different between days (p<0.02 each). The fall in eosinophil numbers was 0.04 (0.09) ×1,000·µL−1 or 35 (50)% at day 8, and 0.07 (0.23) ×1,000·µL−1 or 50 (65)% at day 15. Corresponding reductions in lymphocyte numbers were 0.28 (0.42) ×1,000·µL−1 or 21 (27)% and 0.31 (0.64) ×1,000·µL−1 or 19 (40)%.
When comparing days 1 and 21–29, only the absolute numbers of eosinophils (Wilcoxon, p<0.0001) and the percentage of monocytes (p = 0.015) were significantly different. The fall in eosinophil numbers relative to day 1 was 0.05 (14) ×1,000·µL−1 or 50 (67)%.
The fall in leukocyte and neutrophil numbers at day 15 was also observed in the subgroups of patients who either had or had not received dexamethasone. It reached statistical significance in those without dexamethasone (n = 13; p<0.001 each); only six patients receiving dexamethasone had been able to visit the laboratory at day 15. The changes in monocyte, eosinophil and lymphocyte numbers over the four visits were also statistically significant in the group without dexamethasone (p<0.05 each).
Changes in exhaled H2O2
When comparing days 1, 8 and 21–29, there were no significant changes in the levels of H2O2. In contrast, there was a difference between days 1, 15 and 21–29 (p = 0.040; ANOVA; table 2⇑). The median (interquartile range) fall at day 15 was 0.18 (0.39) µM or 31.0 (56.9)%. There was also a significant difference between days 1, 15 and 21–29 in the patients with COPD of stage II (n = 9) and in the group of the other patients (p = 0.05 each). These two groups did not show a significant difference in baseline H2O2 levels at day 1 (0.46 (0.24) versus 0.70 (0.59) µM).
Changes in FeNO
There was a significant difference in FeNO when comparing days 1, 8 and 21–29 (p = 0.003), but not when comparing days 1, 15 and 21–29 (table 2⇑). Compared with day 1, the level of FeNO was reduced by 3.8 (10.8) ppb or 22.3 (40.2)% at day 8. This fall was also observed in the two subgroups who had (n = 20, p = 0.015) or had not (n = 14, p = 0.037) received dexamethasone, as well as in the subgroup with COPD of stage II (n = 17, p = 0.0014), but not in the other patients.
Variability and correlations
The variables not showing statistically significant differences between days 1 and 21–29 were used to obtain upper limit estimates of variability over the time of the study. Median (interquartile range) values of variability of leukocyte, neutrophil, monocyte and lymphocyte counts, Fe,NO and exhaled H2O2 were 25 (29), 31 (30), 29 (38), 12 (20), 24 (39) and 53 (86)%, respectively. Corresponding intraclass correlation coefficients were 0.81, 0.75, 0.68, 0.77, 0.60 and 0.12.
There were no statistically significant correlations between individual changes of neutrophil numbers and H2O2 levels, or monocyte numbers and FeNO levels. The same was true for leukocyte, eosinophil or lymphocyte numbers, as well as baseline FEV1.
DISCUSSION
The present data indicate that during chemotherapy for lung cancer, the level of H2O2 in EBC decreased in parallel with peripheral blood neutrophil numbers. There was also a fall in the FeNO. This fall occurred prior to the fall in neutrophil numbers and H2O2 levels, at a time when monocyte numbers in peripheral blood were markedly reduced.
The parallel time course of neutrophil numbers and H2O2 concentration, and the discordant course of monocyte, eosinophil and lymphocyte numbers (fig. 1⇑) favour the conclusion that neutrophils were the major source of exhaled H2O2, or at least closely associated with its generation. Conversely, the simultaneous fall of FeNO levels and monocyte numbers suggests monocytes (or macrophages) are basically involved in the production of exhaled NO. It might even be that the recovery of FeNO levels at day 15, when monocyte numbers were still lowered, was linked to the fall in H2O2 level at this time point.
Indeed, a reduction in superoxide production by inhibition of xanthine oxidase was found to be associated with a rise in the level of FeNO 23. An interaction between reactive oxygen species (ROS) and NO is further suggested by the inverse relationship between the levels of exhaled H2O2 and atmospheric inhaled NO 24. Thus, NO production might have been reduced on both day 15 and day 8, but a reduced ROS production allowed more NO to evade into the airway lumen. The fact that FeNO variation was largest at day 15 also appears to be in favour of this explanation. These effects were transient, as FeNO and H2O2 values were back to baseline at days 21–29. The fall in FeNO could also not be explained by dexamethasone, which had been administered in some patients after infusion.
Within the setting of the current study, FeNO was primarily considered as an indicator of scavenging or of potential alterations in the transfer from airway mucosa to lumen. Such alterations could affect the level of exhaled H2O2 in the absence of changes in production. When H2O2 level decreased at day 15, there was no change in FeNO, at least on average. If this was the consequence of less NO being scavenged by ROS, questioning FeNO as indicator of unaltered transfer, the argument still requires a reduction of ROS production, in accordance with the fall in H2O2 level. Taken together with previous evidence that the exhaled H2O2 originates primarily in the airways 17, i.e. the site where the exhaled NO also originates, the present data thus suggest that the fall in H2O2 level at day 15 did indeed represent a decrease of H2O2 production.
Utilising a massive intervention, such as chemotherapy, to study the different effects on different cell types rendered it less difficult to attribute exhaled H2O2 to a specific cell type than comparing patients with marked airway disease would have done. As lenograstim was administered following day 15, it did not affect the effects observed at days 8 and 15. The analysis of the time course of cell differentials confirmed that neutrophils were most affected by chemotherapy. Monocyte, lymphocyte and eosinophil counts also showed reductions, which occurred on both day 8 and day 15, and eosinophil numbers stayed low until days 21–29. These patterns were discordant to the course of H2O2 (fig. 1⇑), whereas FeNO levels fitted closest to monocytes.
As it is known that bronchoalveolar lavage fluid taken from the tumour site contains an elevated number of neutrophils 25, the presence of a tumour might have affected H2O2 levels. However, baseline H2O2 levels of patients (table 1⇑) were not significantly different from those obtained in healthy control subjects measured during the study period (data not shown). There was also no correlation between tumour stage or type and H2O2 level. It is therefore unlikely that local effects, which might be assumed to be particularly sensitive to chemotherapy, have affected the results of this study.
Data obtained in healthy subjects have demonstrated correlations of H2O2 levels with the ability of blood phagocytes or neutrophils to be activated 18, 19, thus indicating the suitability of blood parameters. These studies did not include FeNO measurements, neither did they use an intervention to elicit changes in cell numbers. Neutrophils are known to be abundant in the conducting airways and, given their short lifetime, similar changes in peripheral blood and airways might be expected. In a previous study, however, there was no significant fall in sputum neutrophil counts 21 days after chemotherapy, despite a fall in blood neutrophils 26. Due to restrictions imposed by the patients' therapy schedules, it was not possible to include sputum induction in the present study. The difference might have been due to the variability of sputum cell counts, and the difficulty in drawing conclusions on airway cell numbers from sputum cell density, as well as the timing of measurements. In the present study, measurements were made earlier, close to the expected minimum of cell counts, and patients had less severe airway disease, thus reducing the likelihood for stable disease-related airway neutrophilia. In addition, chemotherapy has probably affected both cell number and activation, and the data regarding blood phagocytes 18, 19 indicate that activation can play a role for exhaled H2O2.
Indeed, using superoxide production as an indicator, a reduction of the neutrophils' ability to produce ROS has been demonstrated for a broad spectrum of chemotherapeutic agents 27. In addition, the platinum compounds used in chemotherapy might have lowered superoxide dismutase activity, as demonstrated in cochlear or renal cells 28, resulting in lower H2O2 production. Conversely, the simultaneous reduction of catalase activity would have reduced H2O2 removal. In contrast to neutrophils, ROS release by macrophages can be enhanced by platinum compounds 29. Thus, chemotherapy might have influenced exhaled H2O2 via ROS production of phagocytes and antioxidant enzyme activities, but the overall effect of these factors in the setting used by the present authors is difficult to estimate. Most importantly, these effects have a rapid onset, in contrast to the delayed H2O2 response observed in the present study.
NO appears to play a role in carcinogenesis and tumour growth via an interplay between reactive nitrogen species and ROS, whereby chemotherapeutic agents exert multiple effects, such as an increase of inducible NO production 30 versus NO-mediated apoptosis 31 in macrophages. Compared with a threshold value of 20 (35) ppb, 24 (7) of the 39 patients showed an elevated level of FeNO, although patients with asthma had been excluded. There was no relationship between FeNO and infections which occurred in six patients, although the highest FeNO value was found in a patient with infection. In addition, blood eosinophils and FeNO were not correlated with each other.
Most patients were ex-smokers who had stopped smoking several weeks before therapy. Smoking is known to lead to a reduction of FeNO levels, probably due to scavenging by ROS. After smoking cessation, FeNO levels rise over ∼1–4 weeks 32. The time of the start of the study, as well as the fall in FeNO level at day 8, did not favour the assumption that smoking cessation has biased FeNO measurements. A similar reasoning applies to H2O2. Smokers without major airway disease show elevated levels of antioxidants and antioxidant enzyme activity in bronchoalveolar lavage fluid 33, but there are no data regarding smoking cessation. Theoretically, a fall in antioxidant levels should have led to an increase in the H2O2 level, not a decrease as observed. Direct measurements of exhaled H2O2 also did not provide evidence that smoking had an effect on H2O2 in mild COPD 3. Thus, it is likely that smoking cessation had no impact on the present findings.
The fact that there was no correlation between individual changes of neutrophil numbers and H2O2 levels was probably due to the large variability of H2O2 measurements, which is known from previous studies 17, 34, 35. As the intervention might have increased variability, coefficients of variation derived by comparing days 1 and 21–29 have to be viewed as upper bounds. Variability of H2O2 was greater than that of other variables, its median value being >50% and nearly twice the magnitude of the fall observed at day 15.
In conclusion, the present data demonstrated a parallel course between the level of exhaled hydrogen peroxide and blood neutrophil counts during chemotherapy. The course of fractional exhaled nitric oxide was different, and was most similar to that of monocyte numbers. On the basis of previous data indicating that the exhaled hydrogen peroxide originates mainly in the bronchi, the present findings suggest that hydrogen peroxide in exhaled breath condensate is primarily a marker of neutrophil number and/or activation, which is detectable via and probably located in the conducting airways.
Acknowledgments
The authors would like to thank all patients for their kind cooperation. They are also grateful to K. Paasch and I. Schwertfeger for technical support, and to A. Bergner, LMU Munich, Germany, for valuable comments.
- Received September 1, 2005.
- Accepted December 23, 2005.
- © ERS Journals Ltd