Usefulness of Quantification and Serial Monitoring of Fine Crackles for Early Detection of Treatment-related Lung Injury: A Report of Two Cases

Abstract

Early detection and appropriate management of treatment-related interstitial lung disease (ILD) are important in cancer treatment. We established an algorithm for quantifying fine crackles using machine learning and reported that the fine crackle quantitative value (FCQV) calculated by this algorithm was more sensitive than chest radiography for detecting interstitial changes. Using this algorithm, we periodically analyzed respiratory sounds in two patients with lung cancer who developed treatment-related ILDs and found that the FCQV was elevated before the diagnosis of ILD. These cases may indicate the usefulness of the FCQV in the early diagnosis of treatment-related ILDs.

Introduction

Thoracic irradiation and antineoplastic drugs, including cytotoxic agents, molecular-targeted agents, and immune checkpoint inhibitors (ICIs), are some of the basic treatment strategies for advanced lung cancer. However, these treatments occasionally cause interstitial lung disease (ILD) as a potentially fatal adverse event. The reported incidence of treatment-related ILD associated with cytotoxic agents and/or ICIs is approximately 1-6% (1-4). In addition, the incidence of radiation pneumonitis associated with radical chemoradiotherapy for lung cancer has been reported to be ≥20% (5,6). Furthermore, treatment-related ILD, especially severe ILD, is associated with poor outcomes (7).

Fine crackles are heard in 60% of patients with interstitial pneumonia and are important physical findings supporting the diagnosis of ILD (8). However, the classification of respiratory sounds has been subjectively performed by physicians, and substantial interobserver disagreement has been reported (9). As a possible solution to these problems, we established numerical values for respiratory sounds and quantitative values for fine crackles (FCQV), course crackles, wheezes, and rhonchi. Using supervised machine learning, we established an algorithm that can output the above four values and the presence or absence of each sound based on the threshold derived from the receiver operating characteristic curve analysis (10). We previously reported that the presence of fine crackles determined by this algorithm was more sensitive than chest radiography in detecting ILD (10).

To evaluate the utility of the FCQV in detecting treatment-related ILD in patients with lung cancer, we recorded and quantified the respiratory sounds of two patients with lung cancer who underwent chemotherapy or chemoradiotherapy. We performed auscultation using dedicated electronic stethoscopes (MSS-U11C) developed by Pioneer (Tokyo, Japan; Supplementary material a), as previously described (9). In brief, we instructed the patients to breathe deeply and then recorded respiratory sounds during at least three breath cycles at six points on the back of each patient (Supplementary material b). The recorded sounds were simultaneously transferred to tablet computers by Bluetooth, and the analyzing software program promptly output the FCQV. In addition, the chronic obstructive pulmonary disease assessment test (CAT), which is considered to be a useful indicator of the health status and prognosis in ILDs, such as idiopathic pulmonary fibrosis (IPF) (11), was recorded at the same time as each auscultation procedure.

Case Reports

Case 1

A 63-year-old man who was a current smoker with a history of chronic obstructive pulmonary disease (COPD) presented to the hospital. Chest computed tomography (CT) showed a 60-mm tumor in the right lower lobe with thickened bronchovascular bundles and ground-glass opacities limited to the subpleural area (Fig. 1a, d). Positron emission tomography-CT (PET-CT) showed an abnormal uptake of ¹⁸F-fluorodeoxyglucose in the right lower lobe tumor and multiple lymph nodes, including at the right neck, bilateral supraclavicular fossae, and hilum of the right lung. He was diagnosed with stage IVA (cT3N3M1b) squamous cell lung cancer and underwent chemotherapy with carboplatin, paclitaxel, ipilimumab, and nivolumab.

Figure 1.

Chest CT scans of the patient in case 1 obtained before the start of chemotherapy (a, d) and on day 12 (b, e). On day 12, ground-glass opacities appeared around the mass shadow. The ground-glass opacities had improved on day 42 after prednisolone treatment (c, f). CT: computed tomography

On day 8, he developed a cough, and his CAT score began to increase, although chest radiography showed no new abnormal shadow (Fig. 2). On day 12, he developed a fever of 38°C, and chest radiography and CT showed expanded ground-glass opacities in both lung fields (Fig. 1b, e, 2). He was diagnosed with drug-induced ILD and administered prednisolone (30 mg/day), resulting in the improvement of both his symptoms and radiological findings (Fig. 1c, f).

Figure 2.

Polygonal lines represent the maximum FCQV in the right lung field (solid line marked with a white circle) and left lung field (solid line marked with a black square) and the COPD assessment test score (dotted line) of the patient in Case 1. The FCQV threshold for determining fine crackles (0.0874) is indicated by a dotted horizontal line. The lower four panels are chest radiography images of the patient in Case 1 taken on day 0 (left-most), day 8 (second from left), day 12 (third from left), and day 33 (right-most). FCQV: fine crackle quantitative value, COPD: chronic obstructive pulmonary disease

The FCQV in his right lung field gradually increased from day 8, peaked on day 16, and then quickly decreased after prednisolone therapy (Fig. 2).

Case 2

An 88-year-old man with a complaint of hemosputum was diagnosed with stage IIIB (cT3N2M0) squamous cell lung cancer. Chest CT showed a 64-mm tumor in the right lower lobe without interstitial abnormalities (Fig. 3a, d). PET-CT showed an abnormal uptake of ¹⁸F-fluorodeoxyglucose in the right pulmonary hilum and mediastinal lymph nodes. The patient underwent radical chemoradiotherapy with weekly carboplatin. Although he developed a cough on day 32 and his CAT score began to increase, chest radiography on day 34 did not show any new lesions (Fig. 4). On day 60, a fever of 38°C was observed, and chest radiography showed ground-glass opacities in the right lower lung field (Fig. 4). Chest CT revealed bilateral ground-glass opacities localized to a previously irradiated field (Fig. 3b, e). He was diagnosed with radiation pneumonitis and administered prednisolone (40 mg/day), resulting in the improvement of ground-glass opacities (Fig. 3c, f).

Figure 3.

Chest CT scans of the patient in Case 2 taken before the start of chemoradiotherapy (a, d) and on day 60 (b, e). On day 60, ground-glass opacities appeared in the bilateral lung fields. The ground-glass opacities had improved on day 113 (c, f). CT: computed tomography

Figure 4.

Polygonal lines represent the maximum FCQV in the right lung field (solid line marked with a white circle) and left lung field (solid line marked with a black square) and the COPD assessment test score (dotted line) of the patient in Case 2. The FCQV threshold for determining fine crackles (0.0874) is indicated by a dotted horizontal line. The lower four panels are chest radiography images of the patient in Case 2 taken on day 0 (left-most), day 34 (second from left), day 60 (third from left), and day 67 (right-most). FCQV: fine crackle quantitative value, COPD: chronic obstructive pulmonary disease

The FCQV in his bilateral lung fields increased markedly from day 30 to 42 and then decreased after the start of prednisolone therapy (Fig. 4).

Discussion

In this report, we presented two cases of treatment-related ILDs that developed in patients with advanced lung cancer who were treated with chemotherapy or chemoradiotherapy. In both cases, respiratory sounds were repeatedly recorded, and the fine crackles were quantified as the FCQV. Importantly, elevation of the FCQV in both cases was found to precede the diagnosis of treatment-related ILD. Through these cases, we can speculate that the periodic analysis of respiratory sounds using our machine-learning-based algorithm can contribute, at least in part, to the early diagnosis of treatment-related ILDs.

Most importantly, this report demonstrated that patients' FCQV increases in parallel with worsening of their symptoms before they are diagnosed with treatment-related ILD. The FCQV represents the fitting degree of the recorded sound to the typical feature of fine crackles. We previously reported that the FCQV was elevated in patients with ILDs, especially in those with IPF (10). In addition, the FCQV was higher in patients with traction bronchiectasis and honeycombing than in those without. In the present two cases, the FCQV was elevated in parallel with the worsening of the patients' symptoms and decreased after treatment initiation. These findings suggest that FCQV can reflect the disease activity and severity of ILDs. In addition, the FCQV remained elevated before the onset of ILD in Case 2. Lung irradiation induces edema in the peripheral airways and alveolar septum (11), which might have led to an increase in the FCQV after starting irradiation.

Recently, the importance of an early diagnosis and appropriate treatment of ILD has been demonstrated for ILD induced by anti-human epidermal growth factor receptor 2 (HER2) antibody-drug conjugate (ADC) (12). As elevation of the FCQV in both cases preceded the diagnosis of treatment-related ILDs, we believe that the serial record and quantification of respiratory sounds may be useful for the early detection of treatment-related ILDs in patients with lung cancer. Of note, the auscultation interval and depth or rate of breath were not standardized in the present cases, and the auscultation points were localized to the back. These limitations might have affected the results. We expect future prospective studies to demonstrate the usefulness of the FCQV in the detection of treatment-related ILDs.

The authors state that they have no Conflict of Interest (COI).

Financial Support

This work was supported by the Japan Agency for Medical Research and Development (AMED).

Supplementary Material

Instruments and auscultation sites

a) Electronic stethoscope and tablet devices, b) Areas where auscultation was performed.