Digital Clubbing Is Associated with Higher Serum KL-6 Levels and Lower Pulmonary Function in Patients with Interstitial Lung Disease

Background Although digital clubbing is a common presentation in patients with interstitial lung disease (ILD), little has been reported regarding its role in assessing patients with ILD. This study evaluated patients with ILD for the presence of clubbing and investigated its association with clinical data. Methods We evaluated patients with ILD who visited the teaching hospital at which the study was conducted, between October 2014 and January 2015. Clubbing, evaluated using a Vernier caliper for individual patients, was defined as a phalangeal depth ratio > 1. We examined the association of clubbing with clinical data. Results Of 102 patients with ILD, we identified 17 (16.7%) with clubbing. The partial pressure of oxygen in arterial blood was lower (65.2 ± 5.9 mmHg versus 80.2 ± 3.1 mmHg; p=0.03), serum Krebs von den Lugen-6 (KL-6) levels were higher (1495.0 ± 277.4 U/mL versus 839.1 ± 70.2 U/mL; p=0.001), and the percent predicted diffusing capacity of carbon monoxide was lower (50.0 ± 6.0 versus 73.5 ± 3.1; p=0.002) in these patients with clubbing. Conclusions Patients with clubbing had lower oxygen levels, higher serum KL-6 levels, and lower pulmonary function than those without clubbing.


Introduction
Interstitial lung disease (ILD) is a heterogeneous cluster of pulmonary disorders characterized by di use parenchymal lung damage [1]. Disease activity in ILD is objectively assessed by pulmonary function testing, chest radiography, and high-resolution computed tomography (HRCT) [1]. Serum Krebs von den Lugen-6 (KL-6) is a biomarker for ILD [2] and can be used to evaluate disease activity in ILD and to predict clinical outcomes [3].
Digital clubbing was rst recognized as a sign of empyema by Hippocrates in the fth century BC [4]. Studied for centuries, clubbing has been associated with a variety of pulmonary diseases, including infections, ILD, and neoplasms [5]. In the clinic setting, the skilled primary care physician and pulmonology specialist may search for clubbing to identify patients with underlying chronic lung disease. However, little has been reported regarding the value of clubbing when assessing patients with ILD. As such, we carried out a cross-sectional assessment of patients with ILD for the presence of clubbing and investigated its association with clinical data, including blood oxygenation measurements, serum KL-6 levels, and pulmonary function test results.

Patients and Study Design.
is cross-sectional retrospective study examined adult outpatients with a diagnosis of ILD visiting the Pulmonology Department at St. Luke's International Hospital (Tokyo, Japan) between October 2014 and January 2015. Patients with a history of diseases shown to be associated with clubbing, including lung carcinoma, asbestosis, active tuberculosis, and sarcoidosis [5], were excluded. Each participant was assessed for clubbing after which the medical charts of participants were reviewed and the following data were extracted: age, sex, pack-years of tobacco smoking, systolic pulmonary arterial pressure (sPAP), and left ventricular ejection fraction (LVEF) estimated by Doppler echocardiography, pulse oximeter oxygen saturation (SpO 2 ), arterial blood gas analysis results, serum KL-6 levels, pulmonary function test results, etiology of ILD, and HRCT imaging patterns of ILD as assessed by radiologists. ILD-GAP index was calculated using the clinical data [6].

De nition of Clubbing.
e presence of clubbing was assessed by objective measurement of a phalangeal depth ratio, calculated by distal phalangeal nger depth (DPD) divided by interphalangeal nger depth (IPD), with the use of a Vernier caliper [7]. Clubbing was de ned as the phalangeal depth ratio greater than 1. While there are many techniques available for assessment of clubbing including pro le sign, hyponychial angle, and Schamroth sign [4], Myers and Farquhar recommend in their systematic review that the phalangeal depth ratio be used for the objective and quantitative assessment of clubbing [4]. A phalangeal depth ratio above 1 is rare in disease-free subjects and was reported to be independent of race, age, and sex [4]. Technically, the caliper was gently placed on an index nger, and distal/interphalangeal nger depth was measured without pressing the tissue [7].

Statistical Analysis.
Categorical variables were compared using Fisher's exact test, continuous variables were compared using the unpaired t-test, and binomial variables were compared using Fisher's exact test. All statistical analyses were performed using SPSS version 21.0 (SPSS Inc., Armonk, NY, USA). A p value < 0.05 was considered to be statistically signi cant.

Ethical Approval.
is study was conducted in accordance with the amended Declaration of Helsinki. Ethical approval was provided by the Ethics Committee of St. Luke's International Hospital, and individual participant consent was obtained.

Study Participants.
A total of 102 patients with HRCTcon rmed ILD met the inclusion criteria, of whom 17 (16.7%) patients had clubbing. Table 1 shows the clinical and demographic characteristics of patients with ILD. Age, sex, pack-years of smoking, sPAP, LVEF, and HRCT imaging patterns were not signi cantly di erent between the two groups.
e diagnostic category of ILD, based on the presence of connective tissue disease (CTD) and the highresolution CT (HRCT) imaging pattern, is shown in Table 2. All patients with connective tissue disease (CTD) had their diagnoses con rmed by an immunology/rheumatology specialist, and a total of 50 patients with connective tissue disease-associated interstitial lung disease (CTD-ILD) were observed. e HRCT imaging pattern was dependent on the radiologists' clinical decision, in accordance with international guidelines [8].

Discussion
In this study, we showed that the presence of clubbing in patients with ILD was associated with lower blood oxygenation levels, higher serum KL-6 levels, and lower pulmonary function when compared to those without clubbing, regardless of the underlying etiology of ILD. Because lower blood oxygenation levels and low pulmonary function are seen in patients with advanced ILD, our results suggest that clubbing may be associated with disease progression. Higher serum KL-6 levels in patients with clubbing also suggest that ILD disease activity may also be associated with the presence of clubbing. In concordance with these ndings, the ILD-GAP score was higher in patients with clubbing, suggesting that the presence of clubbing could also be associated with poor prognosis.
We also found that the prevalence of clubbing in patients with radiological NSIP pattern, RA-ILD, and SSc-ILD was 20.8%, 6.7%, and 25%, respectively. To our knowledge, this study is the rst to report the prevalence of clubbing in SSc-ILD. e prevalence of clubbing in NSIP and RA-ILD was consistent with that in previous reports [9][10][11]. We did not nd any patient with clubbing in those with DM-ILD, which is consistent with previous reports of clubbing being rare in DM-ILD [12]. We con rmed that clubbing is indeed uncommon in DM-ILD; this result would suggest that clubbing in patients with DM-ILD may indicate the presence of an occult pathology, such as lung carcinoma or other CTD. Ishioka et al. reported that four out of 33 patients with CTD-ILD had clubbing, and the prevalence of clubbing in CTD-ILD was signi cantly lower than in patients with idiopathic interstitial pneumonia (16 out of 44 patients) [13]. Similarly, the prevalence of clubbing in patients with CTD-ILD (12.0%) was lower than in patients with non-CTD-ILD (21.1%) in the ere may be speci c cellular or molecular mechanisms underlying CTD-ILD that suppress the emergence of clubbing, or there may be mechanisms in non-CTD-ILD that promote the emergence of clubbing. Further research on the pathophysiology of clubbing is needed to explain the di erence.
We found that hypoxia is associated with clubbing in patients with ILD, and we presume from this that hypoxia could be responsible, to some extent, for clubbing in the study population. Several mechanisms including hypoxia and growth factors have been proposed for the pathogenesis of clubbing; however, the true mechanism of pathogenesis remains elusive [5]. Uppal et al. found that 15-hydroxyprostaglandin dehydrogenase, the enzyme-mediating prostaglandin degradation, is responsible for familial cases of clubbing and hypertrophic osteoarthropathy [14]. ey also showed that elevated prostaglandin E2 (PGE2) levels were found in homozygous familial cases [14]. eir report suggests that clubbing may be mediated by PGE2. In accordance with that theory, Kozak et al. reported that clubbing in patients with lung cancer was associated with elevated urinary levels of PGE2 [15]. Cyclooxygenase-derived prostanoids are reported to be important mediators that regulate pulmonary function in normal and pathological conditions, and with regard to the relationship between PGE2 and ILD, PGE2 has been shown to be a key factor that controls broblast di erentiation and proliferation in mouse models of pulmonary brosis [16]. In addition, cyclooxygenase-2, which induces PGE2, is reported to be widely expressed in the epithelium of patients with idiopathic pulmonary brosis (IPF), asbestosis, or cryptogenic organizing pneumonia [17]. Given these ndings, measuring serum PGE2 levels in patients with ILD who have clubbing might be an interesting area of future research, although other factors, including hypoxia, Note. Data are means ± standard error; SpO 2 � pulse oximeter oxygen saturation; PaO 2 � partial pressure of oxygen in arterial blood; PaCO 2 � partial pressure of carbon dioxide in arterial blood; ILD � interstitial lung disease; KL-6 � Krebs von den Lugen-6; FEV1 � forced expiratory volume in 1 second; FVC � forced vital capacity; VC � vital capacity; DLCO � di using capacity of carbon monoxide; NSIP � nonspeci c interstitial pneumonia; CTD-ILD � connective tissue disease-associated interstitial lung disease; † data were missing for eight patients in the clubbing (−) group for SpO 2 ; 62 patients in the clubbing (−) group and 11 patients in the clubbing (−) group for PaO 2 and PaCO 2 , respectively; three patients in the clubbing (−) group and one patient in the clubbing (+) group for KL-6; 22 patients in the clubbing (−) group and two patients in the clubbing (+) group for %FEV1, %FVC, and %VC; 28 patients in the clubbing (−) group and ve patients in the clubbing (+) group for %DLCO and ILD-GAP index; ‡ data were missing for six patients in the clubbing (−) group for SpO 2 ; one patient in the clubbing (−) group for KL-6; nine patients in the clubbing (−) group for %FEV1, %FVC, and %VC; and 11 patients in the clubbing (−) group for %DLCO and ILD-GAP index; § data were missing for one patient in the clubbing (−) group for KL-6; four patients in the clubbing (−) group for %FEV1, %FVC, and %VC; and one patient in the clubbing (−) group for %DLCO and ILD-GAP index. growth hormones, platelet-derived growth factor, and vascular endothelial growth factor, have also been discussed in relation to the pathogenesis of clubbing [5].
is study has some limitations. First, many data were missing in the pulmonary function tests, which could have biased the results. However, missing data were mostly in the clubbing (−) population, and the patients without pulmonary function test data were clinically stable. at is, the results might have favored the clubbing (+) population. Second, radiographic quanti cation and histopathological assessment of ILD were not performed. In addition, although 52 non-CTD-ILD patients were identi ed in the study, patients with IPF or hypersensitivity pneumonitis were not included.
is was due to di culties in the diagnosis of interstitial pneumonia, the short study period, and the retrospective nature of the study. Lastly, the present study was a singlecenter, cross-sectional study, and the number of outcomes was not enough for adequate power in some analyses.

Conclusions
In conclusion, patients with clubbing showed lower blood oxygenation levels, higher serum KL-6 levels, and lower pulmonary function when compared with patients without clubbing, regardless of the underlying etiology of ILD. e results suggest that clubbing may be associated with disease progression, disease activity, and prognosis in ILD.

Disclosure
An earlier version of this work was presented as a poster at the American oracic Society International Conference, 2016 [18].

Conflicts of Interest
e authors declare that there are no con icts of interest regarding the publication of this article.