Qualitative emphysema and risk of COPD hospitalization in a multicenter CT lung cancer screening cohort study

Abstract

Background:

In the United States, 9 to 10 million Americans are estimated to be eligible for computed tomographic lung cancer screening (CTLS). Those meeting criteria for CTLS are at high-risk for numerous cardio-pulmonary co-morbidities. The objective of this study was to determine the association between qualitative emphysema identified on screening CTs and risk for hospital admission.

Study Design and Methods:

We conducted a retrospective multicenter study from two CTLS cohorts: Lahey Hospital and Medical Center (LHMC) CTLS program, Burlington, MA and Mount Auburn Hospital CTLS program (MAH), Cambridge, MA. CT scans were qualitatively scored by radiologists at time of screening for presence of emphysema. Multivariable Cox regression models were used to evaluate the association between CT qualitative emphysema and all-cause, COPD-related, and pneumonia-related hospital admission

Results:

We included 4675 participants from the LHMC cohort and 915 from the MAH cohort. 57% and 51.9% of the LHMC and MAH cohorts had presence of CT emphysema, respectively. In the LHMC cohort, the presence of emphysema was associated with all-cause hospital admission (HR 1.15, CI 1.03–1.31; p = 0.014) and COPD-related admission (HR 1.64; 95% CI 1.14–2.36; p = 0.007), but not with pneumonia-related admission (HR 1.33; 95% CI 0.96–1.85; p = 0.088). In the MAH cohort, the presence of emphysema was only associated with COPD-related admission (HR 2.05; 95% CI 1.07–3.95; p = 0.031).

Conclusion:

Qualitative CT assessment of emphysema is associated with COPD-related hospital admission. Identification of CT emphysema may provide an opportunity for prevention and early intervention to reduce admission risk.

INTRODUCTION

Chronic obstructive pulmonary disease (COPD) is one of the leading causes of mortality and hospitalization across the world and in the United States.^1–4 Projections estimate that $49 billion will be spent in the United States on medical treatment for COPD in 2020.⁵ The average hospitalization cost for a single episode of COPD exacerbation is approximately $7,100, and the aggregate of hospitalization costs accounts for 45–50% of COPD-related health care expenditures.^6–10 Smoking cessation programs, vaccination against pneumococcal pneumonia, and referral to a pulmonary specialist have all been shown to reduce hospitalizations and improve outcomes in patients with COPD.^11–17 Thus, the ability to identify patients at risk for COPD-related hospitalization is paramount.

Emphysema is one of the pathologic features that defines COPD and can be detected in-vivo using computed tomographic (CT) imaging.¹⁸ Previous investigations have demonstrated that qualitative emphysema assessed on CT is associated with lung cancer, COPD exacerbations, and mortality.^19,20 In the United States, 9 to 10 million patients are estimated to be eligible for CT lung cancer screening (CTLS) based on U.S. Preventive Services Task Force guidelines.²¹ Patients who qualify for CTLS are at high-risk for numerous cardio-pulmonary co-morbidities including COPD. We sought to examine if qualitatively-assessed emphysema, on baseline CTLS scans, could predict all-cause hospitalization and more specifically COPD and pneumonia-related hospitalization.

METHODS

Subjects

This is a retrospective, multi-center study approved by each center’s institutional review board. We included two cohorts in our study, cohort #1 (Lahey Hospital and Medical Center (LHMC), Burlington, MA): all CTLS participants screened between January 1, 2012 and September 30, 2017 (N=4675); and cohort #2 (Mount Auburn Hospital (MAH), Cambridge, MA): all CTLS participants, screened between January 1, 2015 and September 30, 2017, with an in network primary care provider (N=915). To qualify for our study, participants had to satisfy the National Comprehensive Cancer Network (NCCN) Guidelines® Lung Cancer Screening Version 1.2012 high-risk criteria for lung cancer. Based on the NCCN Guidelines®, individuals eligible for lung cancer screening can be classified into NCCN group 1 and 2. This includes group 1) patients aged 55–74 years with ≥ 30 pack-year smoking history and smoking cessation <15 years or group 2) age ≥ 50 years and ≥ 20 pack year smoking history and 1 additional risk factor (other than 2^nd hand smoke). All patients were asymptomatic and had a physician order for CTLS, were free of lung cancer for ≥ 5 years, and had no known metastatic disease. ²²

Clinical Variables

LHMC clinical variables were collected prospectively as part of the CTLS program and stored in a centralized data repository. Additional clinical variables not already available in this data repository were collected retrospectively by manual review of the electronic medical record (EMR) or pulled directly from the EMR and stored using a custom-designed database (FileMaker ProVersion 11; Filemaker Inc, Santa Clara, California). Hospital admissions were collected using Lahey administrative coding data. Principal admission diagnoses of COPD and pneumonia were characterized based on diagnosis codes per 2018 Centers for Medicare and Medicaid Services (CMS) condition-specific measures.²³

MAH clinical variables were collected prospectively as part of the CTLS program and stored in a centralized data repository. Additional clinical variables including COPD and pneumonia hospitalization data were collected retrospectively by manual review of EMR or pulled directly from the EMR. Data was obtained through September 30, 2019 for both cohorts.

Qualitative Emphysema

LHMC:

At the time of screening, staff radiologists visually categorized CT emphysema burden as none, mild, moderate, or marked.

MAH:

At screening, staff radiologists reported CT emphysema as present or absent.

CT scans

LHMC:

All CTLS examinations were performed on ≥64-row multidetector CT scanners (LightSpeed VCT and Discovery VCT [GE Medical Systems, Milwaukee, Wisconsin]; Somatom Definition [Siemens AG, Erlangen, Germany]; iCT [Philips Medical Systems, Andover, Massachusetts]) at 100 kV and 30 to 100 mA, depending on the scanner and the availability of iterative reconstruction software. Axial images were obtained at 1.25–1.5 mm thickness with 50% overlap and reconstructed with both soft tissue and lung kernels. Axial maximum-intensity projections (16 × 2.5 mm) and coronal and sagittal multiplanar reformatted images were reconstructed and used for interpretation.

MAH:

All CTLS examinations were performed on ≥64-row multidetector CT scanners [GE Medical Systems, Milwaukee, Wisconsin]; Siemens FORCE scanner [Siemens AG, Erlangen, Germany]; at 100 kV-150 kV and 40 to 80 mA, depending on the scanner and the availability of iterative reconstruction software. Axial images were obtained at 1.25–1.5 mm thickness with 50% overlap and reconstructed with both soft tissue and lung kernels. Axial maximum-intensity projections (15–16 × 2.5 mm) and coronal and sagittal multiplanar reformatted images were reconstructed and used for interpretation.

Statistical Analysis

Data are presented as means ± standard deviation (SD). Univariate analysis was performed using Cox proportional hazards regression to identify clinical and demographic variables associated with time to first hospitalization (all-cause, COPD-related, and pneumonia-related). Variables known to be associated with risk of admission including age, sex, race, BMI, current smoking status, and pack years of smoking were included in multivariable Cox proportional hazards models for each of the outcomes.²⁴ Additionally, variables associated with hospitalization with p < 0.1 on univariate analyses were also included in the multivariable model. Kaplan-Meier plots were generated to visualize the associations between observed baseline emphysema and all-cause, COPD, and pneumonia admissions. The log-rank test was used to evaluate for significance. Chi-squared tests were used for inter group comparisons. P-values < 0.05 were considered statistically significant. All statistical analyses were performed using STATA14.1 software.

RESULTS

A total of 4,675 participants were screened at LHMC and 1,387 participants were screened at MAH during the study period. Four hundred eighty-seven participants were excluded from the MAH cohort -- 372 were excluded because their primary care providers were out of the MAH network (notably, LHMC participants with out of LHMC primary care providers were not excluded from the LHMC cohort); 100 were excluded because they did not meet criteria for NCCN group 1 or group 2. Of the 4,675 (100%) and 915 (66.0%) included participants from LHMC and MAH, respectively, the mean age was 62.4 ± 6.2 years and 64.4± 6.5. A little over 50% of each cohort was male, and the majority of participants were white. Almost 50% were former smokers. Most participants met criteria for NCCN group 1. Roughly half of the participants had emphysema on baseline CT (Table 1).

Table 1:

Demographics of CTLS Cohorts

	LHMC	MAH
	(N=4,675)	(N=915)
Age	62.4 ± 6.2	64.4 ± 6.5
Sex:
Male	2545 (54.4%)	498 (54.4)
Female	2130 (45.6%)	417 (45.6)
Race:
White Race	4407 (94.3%)	837 (91.5%)
Asian	41 (0.9%)	20 (2.2%)
African American	18 (0.4%)	18 (2.0%)
Other	209 (4.5%)	40 (4.4%)
BMI	29.2 ± 6.0	28.5 ± 6.0
In Network PCP	3226 (69.0%)	915 (100%)
Out of Network PCP	1449 (31.0%)	0 (0%)
Smoking:
Current	2364 (50.6%)	471 (51.5%)
Former	2311 (49.4%)	444 (48.5%)
Pack Years	48.6 ± 21.8	43.4 ± 17.4
Years Quit	10.0 ± 8.4	10.2 ± 9.8
Years Follow Up	3.93 ± 2.14	2.60 ± 1.01
Screening Group:
NCCN Group 1	3658 (78.3%)	773 (84.5%)
NCCN Group 2	1017 (21.8%)	142 (15.5%)
Emphysema
Emphysema (Yes)	2666 (57.0%)	475 (51.9%)
Emphysema (No)	2009 (43.0%)	440 (48.1)
Mild	1839 (39.3%)	-
Moderate	595 (12.7%)	-
Marked	178 (3.8%)	-
Not scored	54 (1.2%)	915 (100%)

There was a similar distribution of emphysema scoring among radiologists for each cohort. At LHMC, three radiologists read the vast majority (3916, 84%) of the scans. Radiologists who read fewer than 100 scans were considered a single radiologist as together they read only 143 (3%) scans. There was a statistically significant difference among radiologists in categorizing no emphysema versus mild emphysema P = <0.001; there was no statistical difference among radiologists for moderate, marked, and unscored emphysema P > 0.05 (Supplemental table 1).

MAH radiologists who read fewer than 100 scans were grouped together; they read a combined total of 228 scans (25% of all scans). Three radiologists reading the remaining 687 (75%) scans. There was no statistical difference among the four radiologist groups in the percentage of participants with emphysema, p > 0.05.

Qualitative Emphysema and All-cause hospital admission

One thousand two hundred forty-eight (26.7%) participants in the LHMC cohort were admitted to the hospital at least once over the course of the study, and there were 2,829 all-cause hospital admissions for this cohort. The hospital admission rates for LHMC and MAH were 387 (8.3%) and 69 (7.5%) at one year, and 679 (14.5%) and 132 (14.4%) at two years, respectively.

LHMC:

Age, pack years, years quit, NCCN group, in network primary care, and presence of emphysema on CT were found to be associated with time to first all-cause hospital admission, Table 2. The effect of emphysema remained significant when adjusted for age, sex, race, BMI, current smoking status, pack years, years quit, NCCN group, and in network primary care (HR 1.15; 95% CI 1.03–1.31; P = 0.014). Figure 1A shows the unadjusted Kaplan-Meier survival plot for presence of emphysema and the probability of being free from all-cause admission over time, P = 0.005.

Table 2:

Cox Univariate Analysis for Time to first All-Cause Hospital Admission

	LHMC		MAH
	HR (95% CI)	P Value	HR (95% CI)
Age	1.02 (1.02–1.03)	<0.001	1.05 (1.03–1.07)	<0.001
Sex (Female)	0.95 (0.85–1.06)	0.330	1.09 (0.82–1.45)	0.551
Race (White)	1.02 (0.75–1.40)	0.880	1.38 (0.75–2.54)	0.296
BMI	1.01 (0.99–1.02)	0.064	0.99 (0.96–1.01)	0.360
Pack Years	1.00 (1.00–1.01)	<0.001	1.00 (0.99–1.01)	0.437
Years Quit	0.99 (0.98–0.99)	0.020	1.01 (0.99–1.02)	0.347
Current Smoker	1.09 (0.97–1.21)	0.150	0.97 (0.73–1.29)	0.844
NCCN group (2)	0.85 (0.74–0.98)	0.024	1.15 (0.80–1.66)	0.456
In Network PCP	1.79 (1.54–2.08)	<0.001	-	-
Emphysema (Yes)	1.18 (1.05–1.32)	0.004	1.34 (1.00–1.79)	0.047
None	Reference Group	-
Mild	1.09 (0.96–1.24)	0.164	ND
Moderate	1.34 (1.13–1.58)	0.001	ND
Severe	1.52 (1.17–1.98)	0.002	ND

Figure 1:

Hospital Admission Kaplan-Meier Plot for Qualitative Emphysema

When emphysema burden was categorized by severity (reference group = no emphysema), moderate and marked emphysema were associated with all-cause hospital admission, as shown in Table 2. These associations remained significant when adjusted for age, sex, race, BMI, current smoking status, pack years, years quit, NCCN group, and in network primary care (HR 1.32; 95% CI 1.11–1.57; P = 0.002 and HR 1.54; 95% CI 1.17–2.03; P = 0.002, respectively). Figure 1B shows the unadjusted Kaplan-Meier survival plot for emphysema severity and the probability of being free from all-cause admission over time, P < 0.001.

MAH:

On univariate analysis, age and the presence of emphysema on CT were found to be associated with time to first all-cause hospital admission, Table 2. The presence of emphysema did not remain significant after adjustment for age, sex, race, BMI, current smoking status, and pack years (HR 1.22; 95% CI 0.90–1.64; P = 0.193). Figure 1C shows the unadjusted Kaplan-Meier survival plot for presence of emphysema and the probability of being free from all-cause admission over time, P < 0.05.

Qualitative Emphysema and COPD admission

One hundred fifty of the 4,675 (3.21%) LHMC and 48 of the 915 (5.3%) MAH patients were admitted for COPD at least once over the course of the study period.

LHMC:

Age, female sex, and CT emphysema were all associated with increased risk of COPD admission as shown in Table 3. The effect of emphysema remained significant after adjusting for age, sex, race, BMI, current smoking status, pack years, years since quitting, NCCN group, and primary care location (HR 1.64; 95% CI 1.14–2.36; P = 0.007). Figure 2A shows the Kaplan-Meier survival plot for emphysema and the probability of being free from COPD admission over time, P <0.002.

Table 3:

Cox Univariate Analysis and COPD admission.

	LHMC		MAH
	HR (95% CI)	P Value	HR (95% CI)	P Value
Age	1.03 (1.01–1.06)	0.018	1.07 (1.03–1.12)	0.003
Sex (Female)	1.54 (1.11–2.12)	0.009	1.56 (0.88–2.77)	0.124
Race (White)	1.62 (0.52–5.09)	0.408	4.05 (0.56–29.39)	0.166
BMI	0.98 (0.95–1.00)	0.101	0.94 (0.89–0.99)	0.035
Pack Years	1.00 (0.99–1.01)	0.253	1.00 (0.99–1.02)	0.535
Years Quit	0.98 (0.96–1.00)	0.107	0.98 (0.94–1.02)	0.299
Current Smoker	1.24 (0.90–1.72)	0.188	0.88 (0.50–1.56)	0.669
NCCN group (2)	0.65 (0.42–1.01)	0.055	0.73 (0.31–1.70)	0.463
PCP Location	1.47 (0.96–2.25)	0.073	-	-
Qualitative Emphysema:	1.74 (1.23–2.47)	0.002	2.50 (1.32–4.72)	0.005
None	Reference		ND
Mild	1.36 (0.92–2.01)	0.125	ND
Moderate	2.17 (1.36–3.46)	0.001	ND
Marked	4.25 (2.43–7.42)	<0.001	ND

Figure 2:

COPD Admission Kaplan-Meir Plot and Emphysema

When emphysema was categorized by severity, both moderate and marked burdens of emphysema, were significantly associated with hospital admission secondary to COPD, as shown in Table 3 The results remained significant so after adjusting for age, sex, race, BMI, current smoking status, pack years, years since quitting, NCCN group, and primary care location, HR 1.98; 95% CI 1.22–3.21; P = 0.006 and HR 3.86; 95% CI 2.14–6.95; P < 0.001. Figure 2B shows the Kaplan-Meier plot for emphysema categorized by severity and the probability of being free from COPD admission over time, P < 0.001.

MAH:

Age and CT emphysema were associated with increased risk of COPD admission as shown in Table 3. The effect of emphysema remained significant after adjusting for age, sex, race, BMI, current smoking status, and pack years, HR 2.05; 95% CI 1.07–3.95; P = 0.031. Figure 2C shows the Kaplan-Meier survival plot for emphysema and the probability of being free from COPD admission over time, P <0.004.

Qualitative Emphysema and Pneumonia admission

One hundred sixty-nine LHMC (3.61%) and 37 MAH (4.0%) patients were admitted for pneumonia at least once over the course of the study period.

LHMC:

Age, in network primary care, and CT emphysema were all associated with increased risk of pneumonia admission as shown in Table 4. After adjustment for age, sex, race, BMI, current smoking status, pack years, and in network primary care there was a non-significant trend toward emphysema being associated with increased risk of pneumonia-related admission, HR 1.33; 95% CI 0.96–1.85; P = 0.088. When categorized by severity, moderate and marked emphysema were significantly associated with increased risk of hospital admission secondary to pneumonia (Figure 3A). On multivariable analysis (adjusted for age, pack years, and in network primary care), only marked emphysema remained significantly associated with risk of pneumonia-related admission, HR 2.45; 95% CI 1.33–4.50; P = 0.005.

Table 4:

Cox Univariate Analysis and Pneumonia admission.

	LHMC		MAH
	HR (95% CI)	P Value	HR (95% CI)	P Value
Age	1.03 (1.01–1.06)	0.011	1.06 (1.01–1.11)	0.014
Sex (Female)	1.13 (0.84–1.53)	0.429	1.25 (0.66–2.39)	0.492
Race (White)	1.02 (0.42–2.48)	0.969	-	-
BMI	0.98 (0.96–1.01)	0.224	0.98 (0.93–1.04)	0.495
Pack Years	1.00 (0.99–1.01)	0.108	1.01 (0.99–1.03)	0.223
Years Quit	0.99 (0.98–1.01)	0.571	0.99 (0.96–1.04)	0.944
Current Smoker	1.08 (0.80–1.46)	0.626	0.81 (0.43–1.55)	0.531
NCCN group (2)	0.85 (0.58–1.24)	0.398	0.59 (0.21–1.67)	0.319
PCP Location	2.09 (1.32–3.30)	0.002	-	-
Qualitative Emphysema:	1.40 (1.02–1.92)	0.036	1.67 (0.85–3.28)	0.137
None	Reference		ND
Mild	1.21 (0.85–1.72)	0.286	ND
Moderate	1.64 (1.05–2.55)	0.030	ND
Severe	2.67 (1.49–4.78)	0.001	ND

Figure 3:

Pneumonia Admission Kaplan-Meir Plot and Emphysema

MAH:

The presence of emphysema on CT was not associated with risk of pneumonia-related admission on univariate analysis. Only age was associated with increased risk of pneumonia admission as shown in Table 4.

Quality Metrics:

In participants with emphysema documented on their baseline CTLS exam and an in network primary care physicians LHMC 1839 (57.0%) and MAH 475 (51.9%). Over 50% of participants with documented emphysema in both cohorts were active smokers at the time of their baseline screening exam. The majority of participants with documented emphysema did not have pulmonary function tests in the 5 years prior to their baseline screening exam and a large % did not have pneumococcal 23 vaccination, Table 5.

Table 5:

Rates of active smoking, PFT/COPD screening in the 5 years prior to baseline exam and pneumococcal 23 (PPSV 23) vaccination rates in patients with qualitative emphysema on their baseline CTLS exam in patients with in network primary care physicians.

	LHMC N = 3226	MAH N = 915
Emphysema (Yes)	1839 (57.0%)	475 (51.9%)
Active Smoking	1045 (56.8%)	261 (54.9%)
PFTs (No)	1,238 (67.3%)	371 (78.1%)
PPSV 23 (No)	713 (38.8%)	254 (53.5%)

DISCUSSION

While several studies have demonstrated an association between CT assessment of emphysema and risk of mortality from lung cancer and COPD, an association between CT emphysema and risk for COPD admission has not been well defined.^25–28 These results also suggest that qualitative emphysema is associated with all cause and pneumonia related hospital admission. However, all case and pneumonia related hospital admission was not replicated in our smaller replication cohort. These results did demonstrate that qualitative assessments of emphysema are associated with an increased risk for COPD admission in both cohorts.

Over half of the patients in our study were active smokers at the time of their baseline scan. This is comparable to previous studies examining CTLS outcomes with current smoking status rates for NLST, NELSON, and MILD at 48.1%, 55.0%, and 77.5%, respectively.^29–31 Studies show that smoking cessation is extremely beneficial for individuals suffering from COPD, regardless of severity.^1,14 For mild and moderate cases of COPD, smoking cessation is associated with lower risk of disease progression, lung cancer, morbidity, and mortality. For severe cases of COPD, smoking cessation is protective against mortality.¹⁴ Smoking cessation has also been shown to reduce the risk of COPD related hospital admission.³² Furthermore, patients referred for combination behavioral and medication assisted smoking cessation programs have much higher quit rates regardless of COPD diagnosis though proper referral to these programs remains variable in primary care settings.^33–35 We have shown that CTLS programs can positively impact smoking cessation rates.³⁶ With effective counseling and adequate resource utilization, we believe that CTLS programs can improve smoking cessation rates and, in turn, improve outcomes for individuals diagnosed with COPD and reduce COPD related hospital admissions.

Spirometry is considered the gold standard for making a clinical diagnosis of COPD given a patient’s symptoms and risk factors. The Global Initiative for Chronic Obstructive Lung Disease’s (GOLD) recommends the use of PFTs for active case finding with symptomatic and high-risk individuals which would include all CTLS eligible patients.^1,37–41 In addition, patients diagnosed with COPD using PFTs are at lower risk of mortality and COPD-related hospital admission.¹³ The NLST-ACRIN sub-study demonstrated that approximately 35% of the screening population has airflow limitation consistent with COPD of which 70% were not previously diagnosed – similar to other studies in high risk populations. ^37,42,43 The majority of patients in both of our cohorts with qualitative emphysema had not received a PFT within five years of their baseline CTLS exam, thus limiting our ability to determine the prevalence of COPD in our cohorts. This observation suggests that there are opportunities to improve COPD screening within CTLS programs, especially among those with emphysema who are at the highest risk for hospitalization.

We also found potential opportunities to improve our vaccination rates especially in patients with documented emphysema. For patients with COPD-related hospitalizations, improvements in pneumococcal vaccination rates are even more critical as they are at an increased risk of mortality.^16,44

As previously reported emphysema is a common incidental finding on CTLS and has the potential to be both a qualitative and quantitative biomarker to help discriminate patients who are at risk for adverse events.^45,46 Prior studies in COPD populations have demonstrated that prior history of COPD admission, more severe airflow obstruction, poorer health status, older age, higher white blood cell count and radiologic evidence of emphysema are associated with the risk of hospitalization for COPD.⁴⁷ The ability to easily identify patients at high risk for hospitalization specifically pneumonia and COPD related hospitalization via CTLS could assist health care institutions in implementing programs that provide targeted clinical care for this subgroup such as pulmonary function testing (PFTs), smoking cessation programs, vaccination against pneumococcal pneumonia and referral to a pulmonary specialist ---- all of which interventions have been proven to reduce hospitalizations and improve outcomes in patients with COPD.^11–17

One of the limitations of our study is that our study populations were predominantly white. This lack of diversity may underestimate the effects of certain social determinants of health on CTLS outcomes. Another limitation is that our hospitalization data only captured in-network hospitalizations, so hospitalizations outside these health systems were likely missed. However, this most likely resulted in an underestimation of the true rates of all cause, pneumonia and COPD hospitalization. The LHMC cohort included patients with both in network and out of network primary care physicians. Finally, the low rates of spirometry in both populations and our lack of hospitalization data prior to the baseline CTLS exam limited our ability to adjust for these known risk factors for COPD admission in our cohorts.

Our results suggest that qualitative emphysema identified on baseline CTLS exams is associated with an increased risk for COPD hospitalization. Further study is needed to identify if qualitative or quantitative scoring of emphysema in CTLS populations can be utilized to develop future models to predict risk for COPD hospitalization and to determine if programs imbedded in CTLS programs that aim to improve COPD screening, immunization and smoking cessation rates can impact risk for COPD admission in this high risk population.

Highlights.

• Qualitative emphysema is associated with risk for COPD hospitalization

• Lung cancer screening represents are an opportunity to screen for COPD

• Lung cancer screening represents an opportunity to improve vaccination rates

• Lung cancer screening represents an opportunity to improve smoking cessation rates