Title: Deterioration of lung diffusion capacity during childhood in sickle cell disease
Abstract: The American Society of Hematology guidelines, 2019, recommended obtaining pulmonary function tests (PFTs) in patients with sickle cell disease (SCD) with various respiratory symptoms even if they are at their steady state.1 These guidelines acknowledged that the usefulness of routine PFT is unknown because of the lack of research. However, this society further suggested that if the PFTs are obtained, it should be a comprehensive study including lung volumes and lung-diffusing capacity for carbon monoxyde (DLCO), in addition to spirometry.1 A large study in adult patients (n = 310) with SCD showed that pulmonary function is abnormal in 90% of adult patients with Hb-SS.2 Common abnormalities included restrictive physiology and decreased DLCO. In this study, decreased DLCO indicated more severe sickle vasculopathy characterized by impaired hepatic and renal function, and a negative linear correlation existed between DLCO and age, suggesting that in adults with Hb-SS, disruption of alveolar–capillary gas exchange progressively deteriorated with time.2 Two recent cross-sectional studies of children with SCD showed that pulmonary function, including DLCO, worsened with age and showed correlations with biological markers of inflammation (induced sputum interleukin-6 levels or blood neutrophilia).3 Overall, these studies highlight the potential interest of DLCO measurement in SCD. The DLCO is the product of KCO (carbon monoxide coefficient of transfer) and alveolar volume (VA) and these two latter indices need to be interpreted separately since the decrease in DLCO is often mitigated by a preserved KCO or even increased KCO in SCD. It has been demonstrated that when corrected for hemoglobin levels, the children with SCD compared to controls of similar age had elevated KCOcorrected. The determination of alveolar-capillary membrane conductance (Dm) and pulmonary capillary blood volume (Vc) from the lung diffusing capacity for carbon monoxide (DLCO) or for nitric oxide (DLNO) has been done in SCD since the seventies, demonstrating an increase in Vc in SCD. KCO is mathematically linked to both Dm and Vc (1/KCO = VA/Dm + VA/θVc); thus, the increase in KCO is related to Vc increase, but since DLCO has been shown to worsen with age, the changes of DLCO, VA, and KCO over time in children with SCD deserve to be studied. The objectives of our study were to describe the evolution of DLCO and its determinants, KCO and VA, and to further assess the initial risk factors of the decrease in DLCO in children/adolescents with SCD. To this end, we retrospectively recorded the routine follow-up PFTs of children with SCD who were included in a prospective cross-sectional study that included the measurement of both DLCO and DLNO with the calculation of Dm and Vc.4 Sixty children/adolescents of Sub-Saharan African or Caribbean ethnicity were prospectively included in the DrepaSympa trial.4 Among them, 43 had subsequent PFTs as part of their routine follow-up (without DLNO measurement). These 43 participants are described in this retrospective study and the 17 nonincluded children are described in the Supporting Information Material. This study was approved by our local Ethics Committee (PHENOBS: N° 2018-430). The parents were informed of the collection of the prospective data for research purposes and they could request that their child be exempted from this study in accordance with French law (non-interventional observational research). First visit: Children underwent spirometry and DLCO/DLNO measurement (Hyp'Air Compact; Medisoft) to calculate the membrane diffusion (Dm) and Vc, as recommended (a finite specific conductance in the blood for NO (θNO) was assumed), as previously described.4 The breath-hold time of this DLCO/DLNO measurement was 4 s. Second visit: Children underwent spirometry and DLCO measurement (without DLNO) according to international recommendations. The breath-hold time of this DLCO measurement was 10 s. The predicted values were those of Global Lung Initiative, taking into account ethnicity for spirometry. Z-scores of DLCO indices were calculated using the reference values of Thomas et al. who provided reference ranges for the outcomes of DLCO/DLNO with a 5 s breath hold, and for the traditional DLCO method, with a 10 s breath hold in Caucasian children.5 The changes in DLCO indices were calculated as visit 2 z-scores minus visit 1 z-scores. The changes in FEV1 were expressed as conditional changes, as recently recommended by ERS/ATS technical standard on interpretive strategies for routine lung function tests. The results were expressed as median (25–75th percentile) since most indices followed a nonnormal distribution. Comparisons between the visits were performed using Wilcoxon signed-rank test. Correlations were evaluated using Pearson's correlation coefficient. Additional statistical analyses are described in the text. A p < .05 was deemed significant. All statistical analyses were performed with StatView 5.0 software (SAS Institute). The characteristics of the 43 children/adolescents at their initial evaluation are described in Table 1 together with their functional characteristics at the final visit. Figure 1 describes the changes in z-scores of DLCO and its components between the two visits and the correlates of DLCOcorrected changes in z-scores. Overall, a significant decrease in DLCO indices is evidenced between the two visits (VA: p < .0001; KCO: p < .0001; KCOcorrected: p < .0001; DLCO: p < .0001; DLCOcorrected: p < .0001) that is also true for spirometry indices with the exception of FEV1/FVC (FEV1: p = .0004; FVC: p = .0011). Nevertheless, the median conditional change score of FEV1 was −1.06 [−1.53; −0.12] and only 6/43 participants (14%) had a significant decrease in FEV1 (conditional change score > −1.96) while the other 37 participants had no significant change in FEV1 (−1.96 ≤ conditional change score ≤ +1.96). The decrease in DLCOcorrected was independently related to both KCOcorrected and VA decreases in a multiple regression (r2 = 0.50; p < .0001). The changes in DLCOcorrected z-scores correlated with DLCOcorrected z-score (Figure 1), VA z-score, Vc z-score (Figure 1), and serum LDH (Figure 1) at the initial visit. In a multiple regression with the change in DLCOcorrected z-score as dependent variable and VA z-score and Vc z-score as independent variables, only Vc z-score remained independently associated with DLCOcorrected z-score change (r2 = 0.27; p = .004). Our original finding is to show that hemoglobin-corrected DLCO indices (DLCO and KCO) deteriorate but from elevated values to near normal values during childhood in SCD. The elevated values of DLCOcorrected in childhood are related to a frank increase in KCOcorrected due to the increase in Vc (all but one z-scores > +1.645). This increase in Vc is at least partly related to the increase in cardiac output due to anemia, thus related to vasodilation that is further limited on exercise. The initial increase in Vc and subsequent DLCOcorrected change correlated; the more the Vc z-score and DLCOcorrected z-score were elevated at baseline (visit 1) and the more was the subsequent decrease in DLCOcorrected. Thus, the deterioration that was evidenced is related to the loss of an adaptive process. Persistent intravascular hemolysis over decades leads to chronic vasculopathy, with ∼10% of patients developing pulmonary hypertension. Thus, follow-up of DLCO may help to detect at risk patients, which warrants further studies. Moreover, DLCO has been correlated with both exertional dyspnea and performance (6 min walked distance) in adult SCD further emphasizing its usefulness.6 A frank deterioration of DLCOcorrected and KCOcorrected z-scores was observed only 2–3 years apart. Nevertheless, at a median 15 years of age, the KCO z-scores were still into the normal range (z-scores −1.645 to +1.645) for 23/43 participants and elevated in the other 20 participants. This result is consistent with the finding of only slightly reduced KCO values (~80% predicted) of young adult patients (~30 years) with SCD.6 Elevated level of serum LDH is a marker of nonspecific tissue damage and its correlation with subsequent lower decrease in DLCOcorrected may seem counterintuitive. Nevertheless, those patients who had the lesser degree of DLCOcorrected decrease over time and the higher levels of LDH were those with normal baseline DLCOcorrected, which may traduce the lack of compensatory vasodilation already present in the patients with more severe disease. Our study has inherent limitations related to its design. Seventeen (28%) participants of the DrepaSympa trial had no follow-up PFTs and were not included; logically they were older explaining their loss of follow-up in a pediatric center. Consequently, our sample (n = 43) was restricted that is a limitation. Finally, DLCO measurement was made with two different breath-hold durations between visit 1 (4 s) and visit 2 (10 s), justifying the use of predicted values that were obtained with similar breath-hold durations.5 Our study has also a strength since we show that the marked deterioration of DLCO and KCO during adolescence in SCD seems to be rather specific of the vascular compartment since only 14% of our participants had a significant decrease in FEV1. In conclusion, our retrospective study shows that DLCO indices significantly decrease in adolescents with SCD from elevated values to almost normal values concerning KCOcorrected and DLCOcorrected, and that the decrease in DLCOcorrected was proportional to the baseline increase in capillary blood volume, while the decrease in DLCO was inversely proportional to baseline LDH concentration. Additional references are provided in the Supplementary material for Introduction, Materials and Methods, and Discussion. Plamen Bokov: Conceptualization; investigation; writing—original draft; methodology; validation. Asmae Zakaria: Investigation; validation; visualization; data curation. Berengere Koehl: Investigation; validation; visualization; data curation; writing—review and editing. Malika Benkerrou: Investigation; validation; visualization; data curation. Christophe Delclaux: Conceptualization; writing—original draft; methodology; formal analysis; project administration; supervision. The authors declare no conflict of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.