Underreporting of Interstitial Lung Abnormalities on Lung Cancer Screening Computed Tomography

To the Editor:

Interstitial lung abnormality (ILA), defined as subclinical bilateral interstitial densities on computed tomography (CT) scan of the chest, are observed in up to 10% of CT scans performed for lung cancer screening (1, 2). Despite evidence that those with ILA are at increased risk of death, hospitalization, and pulmonary function decline (3, 4), current management practices remain uncharacterized. In this multicenter investigation, we assess the reporting of ILA by radiologists reading lung cancer screening CT scans and subsequent management of these patients by primary care physicians.

Methods

This retrospective investigation was conducted at the University of California at Davis and The University of Chicago. Institutional review boards at each institution approved the study and provided a waiver of consent. Radiology databases were used to identify consecutive lung cancer screening CT scans performed from January 1, 2014, to June 1, 2017. A chest radiologist (M.K. or J.H.C.) with interstitial lung disease multidisciplinary experience reviewed lung cancer screening CT scans at each center to identify ILA, defined as bilateral nondependent reticular opacities. CT scans were also assessed for usual interstitial pneumonia (UIP) subtype (typical UIP, probable UIP, indeterminate for UIP, or inconsistent with UIP) (5), interstitial lung disease features (reticulation, honeycombing, traction bronchiectasis, traction bronchiolectasis, and ground-glass opacity), and concurrent emphysema of 10% or greater involvement. ILA with fibrosis was defined as the presence of honeycombing or traction bronchiectasis or bronchiolectasis.

The electronic medical record was retrospectively reviewed to extract pertinent clinical information, including primary care physicians’ documentation and order history that occurred after the lung cancer screening CT scan. Because an aim of the study was to assess pulmonology referral by primary care physicians, patients without an institutional primary care physician and those with an established pulmonologist were excluded. Continuous variables are reported as means with standard deviation. Categorical variables are reported as counts and percentages. The association between ILA reporting by a radiologist and pulmonology referral by a primary care physician was assessed using log-binomial regression, with statistical significance defined as P < 0.05. Statistical analysis was performed using Stata (StataCorp 2013, release 13).

Results

Lung cancer screening CT scans were reviewed for 781 patients, including 364 at University of California at Davis and 417 at The University of Chicago. ILA was detected in 71 (9.1%) cases. Five patients without a primary care physician and seven with an established pulmonologist were excluded, leaving 59 (7.6%) patients included in the final analysis. Cohorts were similar with regard to baseline characteristics, except race distribution, in which white individuals predominated at the University of California at Davis and African Americans at The University of Chicago. The combined cohort mean age was 67.3 years, with 49.2% (n = 29) men and 44.1% (n = 26) current smokers (Table 1).

Table 1.

Clinical characteristics of interstitial lung abnormality cohort

	Not Referred (n = 42)	Referred (n = 17)	Combined (n = 59)
Baseline demographics
Male	20 (47.6)	9 (52.9)	29 (49.2)
Age, yr, mean ± SD	67 ± 6.3	67.5 ± 6.2	67.3 ± 6.2
Race
White	26 (61.9)	7 (41.2)	33 (55.9)
African American	14 (33.3)	9 (52.9)	23 (39)
Hispanic	0 (0)	1 (5.9)	1 (1.7)
Asian	2 (4.8)	0 (0)	2 (3.4)
Smoking history
Current	16 (38.1)	10 (58.8)	26 (44.1)
Past	26 (61.9)	7 (41.2)	33 (55.9)
Supplemental oxygen use	3 (7.1)	1 (5.9)	4 (6.8)
LCS-CT characteristics
ILA reported	27 (64.3)	11 (64.7)	38 (64.4)
Findings only	15 (35.7)	1 (5.9)	16 (27.1)
Findings and impression	12 (28.6)	10 (58.8)	22 (37.3)
ILA with fibrosis	10 (23.8)	9 (52.9)	19 (32.2)
UIP pattern
Typical UIP	0 (0)	3 (17.7)	3 (5.1)
Probable UIP	3 (7.1)	3 (17.7)	6 (10.2)
Indeterminate for UIP	22 (52.4)	5 (29.4)	27 (45.8)
Inconsistent with UIP	17 (40.5)	6 (35.3)	23 (39)
Reticulation	42 (100)	17 (100)	59 (100)
Honeycombing	4 (9.5)	4 (23.5)	8 (13.6)
Traction bronchiectasis	8 (19.1)	5 (29.4)	13 (22)
Traction bronchiolectasis	8 (19.1)	8 (47.1)	16 (27.1)
Ground-glass opacity	12 (28.6)	8 (47.1)	20 (33.9)
Emphysema ≥ 10%	16 (38.1)	9 (52.9)	25 (42.4)
Emphysema reported	14 (87.5)	7 (77.8)	21 (87.5)
PCP characteristics
ILD mentioned in PCP notes	2 (4.8)	5 (29.4)	7 (11.9)
PFT ordered by PCP	3 (7.1)	4 (23.5)	7 (11.9)

Definition of abbreviations: ILA = interstitial lung abnormality; ILD = interstitial lung disease; LCS-CT = lung cancer screening computed tomography; PCP = primary care physician; PFT = pulmonary function test; SD = standard deviation; UIP = usual interstitial pneumonia.

Data presented as n (%) unless otherwise noted.

ILA was noted in 38 of 59 (64%) lung cancer screening CT reports, with 16 of 38 noted in the findings section alone and 22 of 38 noted in the findings and impression sections. Three patients (5.1%) had a typical UIP pattern, 6 (10.2%) had a probable UIP pattern, 27 (45.8) had an indeterminate for UIP pattern, and 23 (39%) had a pattern inconsistent with UIP. Honeycombing was observed in 14% (n = 8) of cases. Concurrent emphysema involving 10% or more of the lungs was observed in 42.5% (n = 25) of cases.

Interstitial lung disease was documented and pulmonary function test ordered by primary care physicians in 12% (n = 7) of cases. A pulmonology referral was placed in 28% (n = 17) of cases. Forty-five percent (10 of 22) of those with ILA mentioned in the lung cancer screening CT report impression received pulmonology referral, compared with 19% (7 of 37) of those without mention (risk ratio, 1.6; 95% confidence interval, 1.0–2.3; P = 0.03). This association persisted after adjusting for center, age, sex, race, smoking history, presence of emphysema, and presence of fibrosis (risk ratio, 1.8; 95% confidence interval, 1.1–2.9; P = 0.02) (Table 2). Active smoking was associated with decreased pulmonology referral in adjusted analysis (risk ratio, 0.3; 95% CI, 0.1–0.7; P = 0.01).

Table 2.

Association between interstitial lung abnormality reporting and pulmonology referral

	ILA Not Mentioned	ILA Mentioned
Referred/total	7/37	10/22
Percent referred (95% CI)	18.9 (8.0–35.2)	45.5 (24.3–67.8)
Unadjusted risk ratio (95% CI)	1 (Reference)	1.6 (1.0–2.3)
Adjusted* risk ratio (95% CI)	1 (Reference)	1.8 (1.1–2.9)

Definition of abbreviations: CI = confidence interval; ILA = interstitial lung abnormality.

^*Adjusted for center, age, sex, race, smoking history, presence of emphysema, and presence of fibrosis.

Discussion

In this investigation, we showed that interstitial lung abnormality was commonly observed on lung cancer screening computed tomography imaging but reported by a radiologist in only 64% of cases. Of those cases with interstitial lung abnormality reported, nearly half were mentioned only in the findings section, which may be missed by clinicians reading only the impression section. In addition, we showed that fewer than 30% of interstitial lung abnormality cases received a pulmonology referral and that reporting of interstitial lung abnormality by a radiologist was associated with a significantly increased likelihood of pulmonology referral. To our knowledge, this study is among the first to explore the practice of interstitial lung abnormality management in an academic setting and underscores the work ahead to improve care for these patients.

Fifteen percent of patients in our cohort had a typical or probable usual interstitial pneumonia pattern, suggesting that a significant minority of patients had undiagnosed idiopathic pulmonary fibrosis (5), assuming known causes of interstitial lung disease were excluded. Idiopathic pulmonary fibrosis is among the most common and deadly forms of interstitial lung disease but now has approved therapy (6), supporting the urgency of early recognition and referral. We also observed that more than 40% of cases had concurrent emphysema. Combined pulmonary fibrosis and emphysema is increasingly recognized as a unique phenotype that may impact outcomes and measures of disease progression (7, 8).

To improve the communication of interstitial lung abnormality observed on lung cancer screening computed tomography to ordering primary care physicians, we suggest standardized use of the Lung-RADS “S” Modifier, which is used to convey significant computed tomography findings unrelated to lung cancer (9). Equally important will be efforts within the pulmonology community to increase awareness among primary care physicians that interstitial lung abnormality warrants evaluation by a pulmonologist. Ideally, such patients would be referred to an interstitial lung disease center of excellence that provides access to a multidisciplinary evaluation and has been linked to improved outcomes (10).

Our study had several limitations. First was the retrospective design, which can only assess association, not causation. Next, despite a multicenter approach, our sample size was small and was limited to patients followed at academic medical centers. In addition, neither center currently has a standardized lung cancer screening program, which may limit the generalizability of our results. A single-reader radiology approach was used in this study, which may have resulted in some false positives, but the consistency of results across centers was reassuring.

Conclusions

Interstitial lung abnormality, increasingly shown to be a clinically significant finding (3, 4), is underreported by radiologists and uncommonly referred to pulmonologists after detection by lung cancer screening computed tomography. Because diagnostic and referral delays are common among patients with interstitial lung disease, computed tomography performed for lung cancer screening serves as a valuable opportunity to improve care for these patients.