Chronic obstructive pulmonary disease and comorbidities’ influence on mortality in non-small cell lung cancer patients

Abstract

Background: In Denmark, lung cancer is the most common cause of cancer-related death and chronic obstructive pulmonary disease (COPD) is the most common comorbidity in patients with non-small cell lung cancer (NSCLC). The aim of this study was to investigate the influence of COPD and other common comorbidities on NSCLC mortality.

Methods: Patients (n = 534) diagnosed with NSCLC at Aalborg University Hospital from 2008–2010 were included retrospectively in this study. Patient records were assessed and the population was dichotomized in COPD and non-COPD subgroups. Comorbidities i.e., ischemic heart disease, hypertension, diabetes mellitus, apoplexia, former malignancy, interstitial lung disease and psychiatric comorbidity were registered and a comorbidity count were calculated. Survival was assessed with log-rank test and uni- and multivariate regression analysis were performed for COPD-status and comorbidity count adjusting for age, gender, BMI, smoking exposure, cancer stage, method of treatment and eastern cooperative cancer group (ECOG) performance status score.

Results: Of 534 NSCLC patients included, 470 were divided into COPD and non-COPD subgroups, 70% with COPD (329/470) and 30% without COPD (141/470). Only 32.5% of the patients in the COPD-group had previously diagnosed COPD. Log-rank test did not show statistically significant difference in survival between the COPD and non-COPD groups (p = .215). Multivariate Cox regression analysis did not show statistically significant association between overall 5-year mortality and the presence of COPD (HR-adj = 0.808, 95% CI = 0.612; 1.068) or other comorbidities (HR-adj = 1.101, 95% CI = 0.979; 1.237) when adjusted for age, BMI, gender, smoking exposure, ECOG performance status, treatment and TNM-stage.

Conclusion: Our findings do not suggest that COPD nor other common comorbidities are significantly associated with higher mortality in NSCLC patients.

Introduction

Chronic obstructive pulmonary disease (COPD) and lung cancer (LC) are two of the most common causes of deaths worldwide, according to the World Health Organization, causing respectively 3.0 and 1.7 million deaths annually [1]. LC causes the most cancer-related deaths of all cancer types worldwide and in Denmark [2] while it is estimated that 430.000 Danes have COPD [3,4]. Smoking is the most important cause of both LC and COPD [5,6]. COPD is the most common comorbidity in non-small cell lung cancer (NSCLC) patients with the prevalence estimated between 50–70% [7,8] and in addition, studies have shown that the majority of NSCLC patients have undiagnosed COPD [8,9].

Several studies have investigated the role of COPD on NSCLC survival but the results are not conclusive. Both Zhai et al. [10] and Sekine et al. [11] have indicated that the prevalence of COPD leads to higher mortality in operable LC patients while Lee et al. [7] and Guillion et al. [12] could not show this association.

While COPD is the focus of several studies on comorbidity and NSCLC survival, it has also been shown that the level of comorbidity is notably high in these populations. Compared to other common cancer types such as breast, colorectal, and prostate cancer it has been indicated that NSCLC populations have the highest comorbidity rates [13]. Ahn et al. [14] have shown that higher comorbidity, quantified as Charlson Comorbidity Index (CCI), is associated with higher mortality rates in patients with NSCLC. In addition to this, Ichina et al. [15] found comorbidities such as cardiovascular disease, cerebrovascular disease, and COPD was associated with higher mortality in NSCLC patients.

The aim of this study was to investigate the influence of (1) COPD on NSCLC mortality and (2) the role of other comorbidities such as diabetes (DM), ischemic heart disease (IHD), hypertension (HT), stroke and former malignancy on NSCLC mortality.

Material and methods

Study population

In this retrospective study, 741 patients were investigated and diagnosed with cancer at Aalborg University Hospital (AaUH) in a 2-year period between 2008 and 2010. Patients were included if registered under diagnosis codes for lung cancer (ICD10 code DR34.0-9) in this time interval. Exclusion criteria were incomplete or missing patient charts, other cancer types than NSCLC or NSCLC cancer relapse. Furthermore, patients with a prior diagnosis of asthma were excluded to prevent the overdiagnosis of COPD. Of the 741 patients assessed 534 patients met inclusion criteria. The inclusion process is depicted in Figure 1.

Figure 1. Flow chart showing selection process of the study population and basis for data analysis.

Data collection and variables

All data were collected from electronic patient case records. Information was registered on gender, age, body mass index (BMI), lung function, eastern cooperative oncology group performance status (ECOG PS), tumor histology, TNM-stage, treatment and survival within 5 years after NSCLC diagnosis.

Active smoking was defined as a current or stopped smoker within 6 months from diagnosis. A former smoker was defined as a person who had stopped smoking more than 6 months prior to NSCLC diagnosis. Smoking exposition was estimated in pack-years [16].

Dynamic lung function parameters i.e., forced expiratory volume within the first second of expiration (FEV1 in liters and % of expected), forced vital capacity (FVC in liters and % of expected), and FEV1/FVC-ratio were gathered. The patients were divided into COPD or non-COPD subgroups. The COPD group consisted of patients with registered COPD and patients with undiagnosed COPD if the FEV1/FVC-ratio < 70%.

Comorbidities such as diabetes, hypertension, ischemic heart disease, stroke, former malignancy, and psychiatric comorbidity (i.e., anxiety, depression) were registered from the patients’ case records. These comorbidities were chosen as they have been shown to be most prevalent in patients with NSCLC [14]. The diagnoses were registered if patients were treated for either of the diseases according to their medical charts. A cumulative comorbidity score i.e., comorbidity count was calculated by the sum total of comorbidities with the exception of COPD.

ECOG PS was registered and systematically rounded down if the score was between two units.

The histological subtypes were classified as adenocarcinoma, squamous cell carcinoma and other from the pathological case records.

The tumor-nodule-metastasis (TNM) stage was registered either from multi-disciplinary conference reports (clinical TNM); pathology reports (pathological TNM). When not apparent in the patient charts, the best estimates of TNM-stage was calculated based on information from pathology reports, computed tomography and positron emission tomography scans by the app Qx Calculate (by Qx MD Software Inc., Vancouver, Canada)

Surgical and oncological treatments were registered under the categories: surgery, chemotherapy and/or -radiation. If none of these modalities were used, the patient was categorized under best palliative care.

Survival was registered in full months from diagnosis to death within the 60 months follow-up.

Statistical analysis

Statistical analysis was performed in SPSS version 25.0. Categorical variables between COPD and non-COPD were compared with χ²-tests and continuous variables were compared with unpaired t-tests. Survival analysis for COPD-status and comorbidity count was conducted with Kaplan-Meier plot and the appertaining log-rank test. Furthermore, to investigate the significance of risk factors on NSCLC survival, uni- and multivariate Cox proportional hazards regression analysis were performed adjusting for age, gender, BMI, COPD-status, comorbidity, ECOG PS, TNM-stage, and treatment. The univariate analyses were also used as feed forward analysis and all factors with p-value < .5 were included in the multivariate analysis. Reported p-values are from two-sided tests and p-values < .05 were considered statistically significant.

The project was registered and approved at the Danish government agency of patient safety (protocol no. 3-3013-1247/1/) and all data were handled according to the regulations of the Danish data protection agency.

Results

Population

A total of 534 patients were included of whom 470 could be categorized in the COPD (329 patients, 70%) and non-COPD (141 patients, 30%) sub groups. Only 32.5% (107/329) of patients in the COPD group were previously diagnosed with COPD.

Table 1 demonstrates the background information on patients, divided into subgroups. The table shows that patients in the COPD-group were significantly older, had higher cumulative tobacco exposure, and were significantly more prone to have squamous cell carcinomas than the non-COPD group.

Table 1. Baseline characteristics of the study population.

	Total		Non-COPD		COPD
n	534		141		329		p-value
Gender
Female	243	45.5%	73	51.8%	138	41.9%	.050
Male	291	54.5%	68	48.2%	191	58.1%
Age (years)	68.4	±9.7	65.9	±10.4	69.4	±9.0	.001*
BMI (kg/m^2)	24.9	±4.9	25.3	±5.3	24.7	±4.8	.284
Smoking status
Never	26	5.0%	15	10.9%	7	2.2%	<.001
Former	191	36.8%	45	32.6%	121	37.5%
Active	302	58.2%	78	56.5%	195	60.4%
Exposure (pack yrs)	37.3	±21.6	32.1	±20.9	40.6	±21.1	<.001
FEV1 (L)	1.78	±0.77	2.12	±0.70	1.62	±0.75	<.001
FEV1 (%)	66.9	±21.6	79.9	±19.2	61.4	±20.3	<.001
FVC (L)	2.82	±0.96	2.82	±0.94	2.82	±0.98	.960
FVC (%)	85.1	±22.2	85.9	±21.7	84.9	±22.7	.681
FEV1/FVC (%)	62.4	±13.3	76.6	±7.9	56.7	±10.3	<.001*
ECOG PS
0-1	417	79.9%	116	84.7%	255	78.9%	.155
≥2	105	20.1%	21	15.3	68	21.1%
Comorbidity
DM	59	11.0%	21	14.9%	34	10.3%	.159
IHD	92	17.2%	23	16.3%	58	17.6%	.729
HT	241	45.1%	59	41.8%	147	44.7%	.570
Apoplexia	38	7.1%	11	7.8%	17	5.2%	.269
Former malignancy	69	12.9%	22	15.6%	40	12.2%	.312
ILD	5	0.9%	2	1.4%	3	0.9%	.624
Psychiatric	65	12.2%	26	18.4%	31	9.4%	.006
Mean no.	1.28	±1.20	1.17	±1.15	1.30	±1.25	.298
Tumor Histology
Adenocarcinoma	221	41.8%	70	50.4%	126	38.4%	.023
Squamous cell carcinoma	196	37.1%	40	28.8%	136	41.5%
Other	112	21.2%	29	20.9%	66	20.1%
Tumor Stage
I	63	12.1%	21	15.1%	36	11.2%	.123
II	100	19.2%	27	19.4%	66	20.5%
III	174	33.4%	36	25.9%	116	36.0%
IV	184	35.3%	55	39.6%	104	32.3%
Treatment
Surgery	164	30.7%	53	37.6%	98	29.8%	.097
Chemotherapy	241	45.1%	65	46.1%	149	45.3%	.872
Radiation	226	42.3%	46	32.6%	153	46.5%	.005
Palliative	112	21.0%	23	16.3%	67	20.4%	.306
Outcome							<.001
Survival (mo.)	21.7	±22.1	24.0	±23.8	21.5	±21.5	.277*
5-y survival (%)	103	19.3%	36	25.5%	58	17.6%	.050

p-value provided for comparison of COPD and non-COPD subgroups.

*Equal variances between the two groups not assumed.

COPD: Chronic obstructive pulmonary disease; BMI: body mass index; FEV1: forced expiratory volume in first second; FVC: forced volume capacity; ECOG PS: Eastern Cooperative Oncology Group performance status; DM: diabetes mellitus; IHD: ischemic heart disease; HT: hypertension; ILD: inflammatory lung disease (ILD).

Furthermore, the COPD group had close to significant lower crude 5-year survival compared to those in the non-COPD group (p = .050). No difference was found in comorbidities between the two groups except a significantly lower prevalence of psychiatric comorbidity in the COPD group.

Figure 2 shows Kaplan-Meier plot comparing cumulative survival for the COPD and non-COPD groups 5 years after diagnosis, with no statistical significance (p = .215). Kaplan-Meier was also performed for cumulative comorbidity, shown in Figure 3, with significant statistical difference (p < .001).

Figure 2. Kaplan-Meier plot for cumulative survival 5 years after diagnosis for the COPD (black) and non-COPD (grey) subgroups (Log-rank test, p = .215).

Figure 3. Kaplan-Meier plot for cumulative survival 5 years after diagnosis for the cumulative comorbidity scores (Log-rank test, p < .001).

Table 2 demonstrates the results of the uni- and multivariate cox regression analysis performed to asses risk factors associated with higher mortality. In the univariate analysis, the presence of COPD was not identified as a statistically significant risk factor. Noticeable risk factors showing statistical significance in this analysis were cumulated comorbidities, performance status >2, TNM-stage, and smoking exposure in pack years.

Table 2. Uni- and multivariate Cox regression performed to asses risk factors associated with NSCLC mortality. Multivariate analysis was performed on 68% (363 patients) of the study population.

		Univariate			Multivariate
	HR	[95% CI]	p-value	HR-adj	[95% CI]	p-value
Age (years)	1.031	1.020–1.041	<.001	1.012	0.997–1.027	.115
Gender
Female	ref.		…	ref.		…
Male	1.232	1.018–1.491	.032	1.306	1.014–1.681	.038
BMI	0.992	0.969–1.014	.461	0.997	0.972–1.024	.851
COPD-status	1.149	0.917–1.440	.226	0.808	0.612–1.068	.134
Smoking (pack yrs)	1.005	1.001–1.010	.025	1.008	1.002–1.013	.009
Comorbidity
Cumulative	1.166	1.075–1.264	<.001	1.101	0.979–1.237	.108
ECOG PS ≥2	1.944	1.545–2.446	<.001	1.149	0.795–1.661	.459
Tumor Stage	2.307	2.051–2.595	<.001	1.910	1.594–2.289	<.001
Treatment
Surgery	0.158	0.121–0.206	<.001	0.261	0.171–0.400	<.001
Chemotherapy	0.926	0.766–1.119	.426	0.727	0.527–1.003	.052
Radiation	1.516	1.250–1.838	<.001	0.791	0.589–1.061	.118
Palliative	3.285	2.621–4.117	<.001	1.632	1.158–2.300	.005

COPD: Chronic obstructive pulmonary disease; BMI: body mass index; ECOG PS: Eastern Cooperative Oncology Group performance status; HR: hazard ratio; HR-adj: adjusted hazard ratio.

The multivariate analysis included 363 patients (68% of the included study population) and was performed to assess the significance of COPD and cumulative comorbidity as risk factors when adjusted for age, gender, BMI, smoking exposure, ECOG PS, TNM-stage, and treatment modalities. COPD-status (HR-adj = 0.808, p = .134) nor cumulative comorbidity (HR-adj = 1.101, p = .108) were identified as risk factors. Statistically significant factors were male gender, smoking exposure, TNM-stage, surgery, and palliative care.

Discussion

Our study shows that COPD and other comorbidities are highly prevalent in NSCLC patients. Only one-third of the COPD patients in this population were diagnosed with COPD, thus COPD was severely underdiagnosed. Although we found that COPD patients had lower 5-year survival, COPD was not associated with higher mortality in either uni- nor multivariate analysis. The presence of common comorbidities was associated with higher mortality in univariate analysis but not in multivariate analysis of NSCLC patients. This discrepancy between the uni- and multivariate analyses is possibly due to over adjustment. Thus, our results do not support our hypotheses that COPD and comorbidity are associated with higher mortality in NSCLC patients.

Several other studies on COPD and NSCLC survival support our findings. Jun Lee et al. [7] found no association in either uni- or multivariate analysis between COPD and 7-year survival in a population of 221 NSCLC patients with similar characteristics to our study population. Lopez-Encuentra et al. [17] investigated a large population of NSCLC patients undergoing surgery and no association between COPD and lower overall 5-year survival could be shown in univariate analysis, although the authors pointed to a difference in conditional survival after 2 years in the conclusion. In contrast to these findings, Kiri et al. [18] reported a lower 3-year overall survival in patients with prior registered COPD in univariate analysis. This was a database study consisting of a large population of patients with all types of lung cancer and only a very small proportion were registered as suffering from COPD. These findings were supported by Zhai et al. [10] who concluded that patients with COPD and lung cancer had higher mortality. In this study, the COPD-diagnosis was self-reported and only lower stage (IA–IIB), surgically resected lung cancer patients were included. Sekine et al. [11] found no difference in overall survival between COPD and non-COPD patients but lower intercurrent survival in the COPD group in univariate analysis.

Although highly significant in univariate analysis, we found no association between total number of comorbidities and mortality in a multivariate analysis in this study. The validity of this result is supported by the other findings in the multivariate analysis – notably ECOG performance status is not associated with mortality while tumor stage, being most the important indicator of radical treatment, is highly significant. Our result is inconsistent with the findings of Ahn et al. [14], which suggests that comorbidity expressed in Charlson Comorbidity Index(CCI) was associated to lower survival in stage I-III NSCLC patients. Tammemagi et al. [13] also found comorbidity expressed in both CCI and comorbidity count was associated with higher mortality in patients with all lung cancer types. Iachina et al. [15] showed in a large register-based study on Danish NSCLC patients that COPD, cardiovascular and cerebrovascular disease, and diabetes are associated with higher mortality in multivariate analysis.

The existing research in the field of NSCLC survival and comorbidity is mostly retrospective, and tend to indicate that more severe degrees of COPD and higher comorbidity burden, in general, is associated with lower survival. This effect seems more detectable in surgical patients and with lower stages of NSCLC disease, although our findings do not support this. The reason for this divergence could be due to the inclusion of all stages of NSCLC and a relatively small study population in this study. Further prospective research is required in this field.

The limitations of our study are mainly due to the retrospective study design using clinical records as data resource, which inevitably results in a number of missing data. Furthermore, the quantification of comorbidity poses a challenge. In this study, we cumulated the most common comorbidities while other studies have used various comorbidity index scores e.g., CCI. The inclusion of not previously diagnosed COPD patients due to obstructive lung function does of course include risk of having included undiagnosed asthma patients in this population. However, the number of never-smokers was limited in the COPD population and we, therefore, assume that the risk of bias from potential undiagnosed asthma patients is limited.

Notable strengths in this study are reliable and consistent information in the patient charts and exact information on survival.

Conclusion

This study suggests that neither COPD nor other common comorbidities have any impact on survival among patients with NSCLC in all disease stages and should therefore not restrict treatment strategies. However, prospective studies with larger study populations are required in this research field.

Disclosure statement

No potential conflict of interest was reported by the authors.