The Glasgow Prognostic Score Predicts Survival in Patients with Advanced Non-Small Cell Lung Cancer Harboring Sensitive Epidermal Growth Factor Receptor Mutations Who Are Treated with Epidermal Growth Factor Receptor-Tyrosine Kinase Inhibitors

Abstract

Introduction

Epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitors (TKIs) are the standard first-line treatment for advanced non-small cell lung cancer (NSCLC) with sensitive EGFR mutations. The Glasgow prognostic score (GPS) is an inflammation-assessing score based on C-reactive protein and albumin concentrations. Information regarding the association between the GPS and EGFR-TKI treatment effectiveness is limited; hence, we investigated whether the GPS can predict the response of NSCLC to EGFR-TKIs.

Methods

We evaluated 340 patients with NSCLC harboring sensitive EGFR mutations who received EGFR-TKI monotherapy between March 2009 and July 2021. The Kaplan-Meier method and Cox proportional hazards models were used to assess progression-free survival (PFS) and overall survival (OS).

Results

After a median follow-up of 26.6 months, patients with a GPS of 0, 1, and 2 had PFS of 15.7, 10.0, and 6.3 months, respectively, and OS of 40.1, 25.8, and 14.4 months, respectively; patients with a GPS of 0 had significantly better PFS and OS than those with a GPS of 1 (p = 0.03, p = 0.001, respectively) or 2 (p < 0.001, p < 0.001, respectively). Multivariate analysis identified poor performance status, stage 4 at diagnosis, type of EGFR-TKI (gefitinib/erlotinib vs. afatinib), and GPS = 2 as predictors of a short PFS. Meanwhile, poor performance status, gefitinib/erlotinib administration, and GPS = 2 were predictors of a short OS.

Conclusion

The GPS predicted the survival of NSCLC patients harboring sensitive EGFR mutations who were undergoing EGFR-TKI treatment. The GPS might be ideal for routine use in clinical practice, given that it is an easily calculated parameter.

Introduction

Lung cancer is one of the most serious malignancies worldwide [1]. Non-small cell lung cancer (NSCLC) accounts for 85% of all lung cancers [2], with most patients diagnosed at an advanced stage. Patients with NSCLC frequently experience weight loss and a systemic inflammatory response, which can lead to cancer-associated cachexia [3, 4]. The Glasgow prognostic score (GPS) was initially developed by Forrest et al. [5] as a prognostic indicator for patients with advanced NSCLC; subsequently, various studies found that the GPS is an independent predictor of survival in patients with various cancers, including NSCLC [6–11]. The GPS is an evaluation system that relies on serum C-reactive protein (CRP) and albumin levels to assess systemic inflammatory response. In this context, CRP is a nonspecific inflammatory marker that can also be used to identify poor nutritional status and assess the risk of poor overall survival (OS) [12, 13], while albumin is a nutritional marker that is inversely correlated with CRP [14]. Previous studies revealed that the GPS can be used to predict the prognosis of patients with NSCLC who are undergoing cytotoxic chemotherapy [15, 16]. Additionally, the GPS was recently reported to be a robust predictor of progression-free survival (PFS) and OS in patients with NSCLC who were treated with immune checkpoint inhibitors [17, 18]. Mutations in the gene that encodes the epidermal growth factor receptor (EGFR) are key drivers of NSCLC, and patients with such mutations can be effectively treated using EGFR-tyrosine kinase inhibitors (TKIs) such as gefitinib, erlotinib, afatinib, and osimertinib [19–24]. However, there is limited information regarding the association between the GPS and EGFR-TKI treatment effectiveness. Therefore, this study investigated whether the GPS could help predict the efficacy of first-line EGFR-TKI treatment in patients with EGFR-mutated NSCLC.

Patients and Methods

Study Design and Patient Selection

This retrospective observational cohort study was conducted at Kitasato University Hospital between March 2009 and July 2021. The subjects were patients with advanced NSCLC who received EGFR-TKI monotherapy, including gefitinib, erlotinib, afatinib, and osimertinib, as a first-line treatment. The inclusion criteria were as follows: (1) histologically or cytologically confirmed NSCLC harboring either an exon 19 deletion or exon 21 L858R mutation in the EGFR gene, (2) stage 4 disease, (3) postoperative recurrence according to the new Union for International Cancer Control criteria (version 8), (4) at least one measurable lesion for evaluating disease control or progression, and (5) the ability to receive oral treatment.

Data on patient characteristics, including age at diagnosis, sex, Eastern Cooperative Oncology Group (ECOG) performance status (PS) at the start of EGFR-TKI treatment, smoking status, clinical stage, tumor histology, brain metastasis status, EGFR alteration, type of EGFR-TKI, and laboratory data, including CRP and albumin levels at the start of EGFR-TKI treatment, were collected from medical charts. This study was approved by the ethical review board of Kitasato University and its affiliated hospitals (approval number B21-095); the board permitted the use of the opt-out method in lieu of written informed consent.

GPS Evaluation and Analysis of EGFR Mutations

Serum CRP and albumin levels were measured either on the day of the initiation of TKI treatment or 1 day before. The GPS was categorized into three groups as follows: a GPS of 0 denoted CRP <1.0 mg/dL in conjunction with an albumin level ≥3.5 g/dL, a score of 1 denoted either CRP elevation or albumin decrease alone, and a score of 2 indicated CRP ≥1.0 mg/dL in conjunction with an albumin level <3.5 g/dL. Initial testing for EGFR mutation status was performed using a sample of the primary tumor, a metastatic lesion, or pleural effusion fluid via the peptide nucleic acid-locked nucleic acid polymerase chain reaction clamp, Cobas method, or Oncomine Dx Target Test.

EGFR-TKI Treatment

All patients received a single daily starting dose of gefitinib (250 mg), erlotinib (150 mg), afatinib (30 or 40 mg), or osimertinib (80 mg). The starting dose of afatinib was determined by the physician-in-charge according to the patient’s background, with 14 and 36 patients receiving 30 mg and 40 mg, respectively.

Statistical Analysis

We used a χ² test to evaluate baseline characteristics according to GPS status. PFS was defined as the interval between EGFR-TKI initiation and disease progression or death. OS was defined as the interval between the initiation of EGFR-TKI and death; patients who were alive were censored on the date of the last follow-up visit. The Kaplan-Meier method was used to estimate survival, with differences analyzed using log-rank test. Cox proportional hazards models with stepwise regression were applied to identify factors predictive of PFS and OS; the variables included sex, age, PS, smoking status, clinical stage at diagnosis, status of brain metastasis, tumor histology, EGFR alteration, type of EGFR-TKI, and GPS. The results are presented as hazard ratios (HRs) and 95% confidence intervals (CIs). Statistical significance was set at a p value ≤ 0.05. All statistical analyses were performed using SPSS software version 28.0 for Windows (IBM Corp., Armonk, NY, USA).

Results

Patient Characteristics

As shown in Figure 1, a total of 340 patients with advanced NSCLC harboring sensitive EGFR mutations were included in the final analysis; their basic characteristics are shown in Table 1. Sixty percent of the patients were women, and the overall median age was 70 years (range, 37–90 years). Moreover, 74% had an ECOG PS score of 0 or 1, 24% had brain metastasis, 52% had an exon 19 deletion, and 48% had an L858R point mutation. In total, 66%, 16%, and 17% of the patients had a GPS of 0, 1, and 2, respectively. The ECOG PS scores, smoking status, and clinical stage at diagnosis were significantly correlated with the GPS (Table 2).

Fig. 1.

Flowchart depicting patient selection. EGFR, epidermal growth factor receptor; NSCLC, non-small cell lung cancer; TKI, tyrosine kinase inhibitor.

Table 1.

Characteristics of the 340 patients included in the study

	N (%)
Sex
Female	205 (60)
Male	135 (40)
Age, years, median (range)	70 (37–90)
ECOG PS score
0–1	253 (74)
2–3	87 (26)
Smoking status
Never smoker	206 (61)
Current or former smoker	134 (39)
Stage
IV	249 (73)
Postoperative recurrence	91 (27)
Brain metastasis
Negative	258 (76)
Positive	82 (24)
Histology
Adenocarcinoma	315 (93)
Other	25 (7)
EGFR alteration
Exon 19 deletion	177 (52)
L858R point mutation	163 (48)
EGFR-TKI
Gefitinib/erlotinib	202 (59)
Afatinib	49 (14)
Osimertinib	89 (27)
Laboratory data, median (range)
CRP, mg/dL	0.27 (0.03–18.9)
Albumin, g/dL	3.9 (1.6–5.0)
GPS
0	226 (66)
1	56 (16)
2	58 (17)

ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitor.

Table 2.

Association of various patient-related factors with the GPS

	GPS 0	GPS 1	GPS 2	p valuea
	N = 226	N = 56	N = 58
Sex
Female	136	33	36	0.94
Male	90	23	22
Age, years
<75	147	36	36	0.92
≥75	79	20	22
ECOG PS score
0–1	193	28	32	<0.001
2–3	33	28	26
Smoking status
Current or former smoker	75	25	34	0.001
Never smoker	151	31	24
Stage
IV	153	42	54	<0.001
Postoperative recurrence	73	14	4
Brain metastasis
Positive	50	18	14	0.29
Negative	176	38	44
Histology
Adenocarcinoma	211	51	53	0.78
Other	15	5	5
EGFR alteration
Exon 19 deletion	115	29	33	0.72
L858R point mutation	111	27	25
EGFR-TKI administered
Osimertinib	58	18	13	0.40
Afatinib	36	8	5
Gefitinib/erlotinib	132	30	40

GPS, Glasgow prognostic score; ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; TKI, tyrosine kinase inhibitor.

^aχ² test.

Survival Analysis

The cut-off date for survival analysis was July 2022, and the median follow-up period for all patients was 26.6 months (range, 0.8–71.6 months). The median follow-up periods among patients who received gefitinib/erlotinib, afatinib, and osimertinib were 25.9 (range, 0.2–136.1) months, 37.2 (range, 0.8–71.6) months, and 24.1 (range, 3.2–45.4) months, respectively. Per Kaplan-Meier analysis, the median PFS was 15.7 months (95% CI: 13.9–17.5) in the GPS = 0 group, 10.0 months (95% CI: 7.2–12.8) in the GPS = 1 group, and 6.3 months (95% CI: 2.4–10.2) in the GPS = 2 group (Fig. 2); patients with a GPS of 0 had significantly more favorable PFS than those with a GPS of 1 (p = 0.03) or 2 (p < 0.001). The median OS was 40.1 months (95% CI: 35.0–45.2) in the GPS = 0 group, 25.8 months (95% CI: 11.0–40.6) in the GPS = 1 group, and 14.4 months (95% CI: 8.2–20.6) in the GPS = 2 group; again, patients with a GPS of 0 had significantly longer OS than those with a GPS of 1 (p = 0.001) or 2 (p < 0.001) (Fig. 3).

Fig. 2.

Kaplan-Meier plots showing progression-free survival (PFS) among patients categorized using the Glasgow prognostic score (GPS). CI, confidence interval.

Fig. 3.

Kaplan-Meier plots showing overall survival (OS) among patients categorized using the Glasgow prognostic score (GPS). CI, confidence interval.

On univariate analysis, a poor PS score of 2 or 3 (HR 2.36, p < 0.001), gefitinib/erlotinib treatment (HR 2.03, p < 0.001), a GPS of 1 (HR 1.44, p = 0.034), and a GPS of 2 (HR 1.40, p < 0.001) were associated with a significantly shorter PFS, whereas postoperative recurrence (HR 0.53, p < 0.001) and the lack of brain metastasis (HR 0.68, p = 0.006) were significantly associated with a longer PFS (Table 3). On multivariate analysis, a poor PS score (HR 2.00, p < 0.001), afatinib treatment (HR 1.79, p = 0.01), gefitinib/erlotinib treatment (HR 2.40, p < 0.001), and a GPS score of 2 (HR 1.91, p < 0.001) were associated with a significantly shorter PFS, whereas postoperative recurrence (HR 0.69, p = 0.033) was significantly associated with a longer PFS (Table 3).

Table 3.

Cox proportional hazards analyses of factors potentially associated with PFS

variable	Univariate analysis		Multivariate analysis
	HR (95% CI)	p value	HR (95% CI)	p value
Sex
Male	1 (ref.)	0.34	1 (ref.)	0.68
Female	0.88 (0.769–1.14)		0.94 (0.70–1.27)
Age, years
<75	1 (ref.)	0.48	1 (ref.)	0.99
≥75	1.10 (0.85–1.41)		1.002 (0.76–1.32)
ECOG PS score
0–1	1 (ref.)	<0.001	1 (ref.)	<0.001
2–3	2.36 (1.80–3.10)		2.00 (1.45–2.70)
Smoking status
Current or former smoker	1 (ref.)	0.07	1 (ref.)	0.085
Never smoker	0.79 (0.62–1.02)		0.76 (0.55–1.04)
Stage
IV	1 (ref.)	<0.001	1 (ref.)	0.033
Postoperative recurrence	0.53 (0.39–0.70)		0.69 (0.49–0.97)
Brain metastasis
Positive	1 (ref.)	0.006	1 (ref.)	0.16
Negative	0.68 (0.51–0.89)		0.80 (0.59–1.09)
Histology
Adenocarcinoma	1 (ref.)	0.051	1 (ref.)	0.10
Other	1.61 (1.00–2.62)		1.62 (0.91–2.88)
EGFR alteration
Exon 19 deletion	1 (ref.)	0.16	1 (ref.)	0.082
L858R point mutation	1.19 (0.93–1.52)		1.25 (0.97–1.61)
EGFR-TKI
Osimertinib	1 (ref.)		1 (ref.)
Afatinib	1.30 (0.93–1.81)	0.13	1.79 (1.15–2.78)	0.01
Gefitinib/erlotinib	2.03 (1.47–2.82)	<0.001	2.40 (1.70–3.39)	<0.001
GPS
0	1 (ref.)		1 (ref.)
1	1.44 (1.03–2.01)	0.034	1.21 (0.84–1.72)	0.30
2	1.40 (1.20–1.64)	<0.001	1.91 (1.37–2.67)	<0.001

CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; GPS, Glasgow prognostic score; TKI, tyrosine kinase inhibitor.

With respect to OS, univariate analysis revealed that a poor PS score (HR 3.05, p < 0.001), lack of brain metastases (HR 0.67, p = 0.0011), gefitinib/erlotinib treatment (HR 1.50, p < 0.001), a GPS of 1 (HR 1.82, p = 0.001), and a GPS of 2 (HR 1.58, p < 0.001) were associated with a significantly poorer OS, whereas postoperative recurrence (HR 0.56, p = 0.001) was associated with a significantly longer OS (Table 4). On multivariate analysis, a poor PS score (HR 2.22, p < 0.001), gefitinib/erlotinib treatment (HR 2.50, p < 0.001), and a GPS of 2 (HR 2.03, p < 0.001) were associated with significantly shorter OS (Table 4).

Table 4.

Cox proportional hazards analyses of factors potentially associated with OS

variable	Univariate analysis		Multivariate analysis
	HR (95% CI)	p value	HR (95% CI)	p value
Sex
Male	1 (ref.)	0.26	1 (ref.)	0.57
Female	0.85 (0.65–1.12)		0.91 (0.65–1.28)
Age, years
<75	1 (ref.)	0.41	1 (ref.)	0.36
≥75	1.13 (0.85–1.50)		1.15 (0.85–1.57)
ECOG PS score
0–1	1 (ref.)	<0.001	1 (ref.)	<0.001
2–3	3.05 (2.30–4.05)		2.22 (1.54–3.09)
Smoking status
Current or former smoker	1 (ref.)	0.07	1 (ref.)	0.026
Never smoker	0.77 (0.59–1.02)		0.66 (0.46–0.95)
Stage
IV	1 (ref.)	0.001	1 (ref.)	0.15
Postoperative recurrence	0.56 (0.40–0.78)		0.75 (0.50–1.11)
Brain metastasis
Positive	1 (ref.)	0.011	1 (ref.)	0.66
Negative	0.67 (0.49–0.91)		0.92 (0.65–1.32)
Histology
Adenocarcinoma	1 (ref.)	0.61	1 (ref.)	0.69
Other	1.19 (0.71–2.02)		1.14 (0.61–2.13)
EGFR alteration
Exon 19 deletion	1 (Ref.)	0.13	1 (Ref.)	0.11
L858R point mutation	1.23 (0.94–1.61)		1.26 (0.95–1.67)
EGFR-TKI
Osimertinib	1 (ref.)		1 (ref.)
Afatinib	1.56 (0.81–2.99)	0.18	1.28 (0.71–2.33)	0.41
Gefitinib/erlotinib	1.50 (1.20–1.87)	<0.001	2.50 (1.61–3.87)	<0.001
GPS
0	1 (ref.)		1 (ref.)
1	1.82 (1.26–2.63)	0.001	1.25 (0.83–1.87)	0.29
2	1.58 (1.34–1.87)	<0.001	2.03 (1.42–2.90)	<0.001

CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; GPS, Glasgow prognostic score; TKI, tyrosine kinase inhibitor.

The PFS and OS, according to the GPS (0–1 vs. 2), were analyzed separately among patients with poor PS scores (2–3). The PFS of those with a GPS of 0–1 and 2 were 6.7 months (95% CI: 5.6–7.8) and 1.9 months (95% CI: 0.3–3.5), respectively; the former group tended to have a longer PFS than the latter (p = 0.07) (Fig. 4a). Moreover, the OS of patients with a GPS of 0–1 and those with a GPS of 2 were 14.4 months (95% CI: 9.0–19.8) and 3.4 months (95% CI: 0.0–7.4), respectively; this difference was significant (p = 0.008) (Fig. 4b).

Fig. 4.

Kaplan-Meier plots showing survival in patients with poor Eastern Oncology Group performance status scores (2–3) categorized using the Glasgow prognostic score (GPS). a Progression-free survival (PFS). b Overall survival (OS). CI, confidence interval.

Discussion

We found that a high GPS (i.e., a score of 2 denoting CRP ≥1.0 mg/dL and albumin <3.5 g/dL) before the initiation of EGFR-TKI treatment was significantly associated with early disease progression and death in patients with advanced EGFR-mutated NSCLC. Previously, Kasahara et al. [25] reported that the GPS predicted the efficacy of EGFR-TKI treatment (gefitinib, erlotinib, and afatinib) among patients with NSCLC who had EGFR mutations. The findings of our study were consistent with theirs; however, ours is the first study (to our knowledge) that assessed the relationship between the GPS and the effectiveness of EGFR-TKI including the third-generation EGFR-TKI, osimertinib.

We found that the GPS was significantly correlated with the ECOG PS score and clinical stage. Additionally, the GPS was found to be an independent predictor of PFS and OS, as well as of clinical stage and ECOG PS score. Postoperative recurrence has previously been identified as a favorable prognostic factor compared with stage 4 disease in patients with EGFR-mutated NSCLC who received EGFR-TKIs [26, 27]; as such, our findings in this respect are consistent with existing data. While the reasons for this observation remain unclear, we propose two hypotheses. First, routine follow-up using imaging modalities after curative surgery can detect early disease recurrence; accordingly, the tumor burden in patients with recurrence is usually lower than that in patients with stage 4 NSCLC at diagnosis. In support of this reasoning, we previously found that the number of metastatic lesions among patients with recurrent NSCLC harboring sensitive EGFR mutations was lower than that among patients with stage 4 disease [26]. Second, given the higher possibility of tumor heterogeneity in patients with a large tumor burden, any administered anticancer drugs are likely to be less effective [28, 29]. Taken together, these differences in both tumor burden and heterogeneity between patients with stage 4 disease and those who experience postoperative recurrence may explain the favorable PFS and OS in the latter group [30, 31].

PS has previously been identified as a potent prognostic factor in patients with lung cancer [32, 33]. Our data provided similar evidence, suggesting that our study population is representative of patients with NSCLC in general. The ECOG PS is a subjective index scoring system that is useful for evaluating patients’ general well-being; in contrast, the GPS is an objective index scoring system that can be used to classify patients based on their combined albumin and CRP levels. The GPS is derived from serum protein concentrations that can be obtained during routine inpatient blood sampling and is thus easily available; accordingly, it can be readily adopted as part of clinical practice at most institutions. Notably, it was previously reported that the GPS had a greater prognostic value than that of the ECOG PS score [34, 35]. Whereas EGFR-TKIs are key therapies for patients with EGFR-mutated NSCLC, data regarding their effectiveness in patients with a poor PS are scarce [36–38]. In the current study, we found that patients with a poor PS had worse PFS and OS if their GPS values were 2; this might help determine the most suitable TKI for this subgroup. Taken together, it is clear that the two scoring systems (PS and GPS) can complement each other in terms of predicting survival; therefore, it is reasonable to consider both the GPS and PS together when identifying patients with EGFR-mutant NSCLC who are more likely to derive a clinical benefit from EGFR-TKIs.

Previous studies found that the body mass index (BMI) is associated with nutritional status and cachexia and that it is a prognostic factor for patients with various malignancies [39–41]. However, Yun et al. [42] found that the BMI was not a statistically significant predictor of either PFS or OS in patients with EGFR-mutated NSCLC who were treated with EGFR-TKIs. Given that the GPS is based on both albumin and CRP, it is indirectly indicative of weight and muscle loss, drug metabolism, cytokine level elevation, and adipokine levels [4, 43–47]. As such, the GPS reflects a patient’s nutritional status as well as BMI. Our data indicate that the GPS contrasts with the BMI, given the former’s ability to be used as a predictor of survival following EGFR-TKI treatment.

This study had several limitations. First, it was a retrospective analysis of a heterogeneous patient cohort and follow-up data; as such, our results should be interpreted with caution, given the potential bias in our data. Second, the follow-up time for evaluating OS among patients who received osimertinib was relatively short, as this agent was only approved for use in Japan in 2018. Third, BMI data were not included in this analysis, given that it was not consistently measured before commencing EGFR-TKI treatment.

In conclusion, we found that the GPS predicted the survival outcomes of patients with NSCLC who had sensitive EGFR mutations and who received EGFR-TKIs. Although further studies are required to validate our findings, the GPS is an easily calculated parameter and might therefore be ideal for routine use in clinical practice.

Acknowledgments

We are grateful to the staff members of the Department of Respiratory Medicine, Kitasato University School of Medicine, for their suggestions and assistance.

Statement of Ethics

All procedures performed in this study were in accordance with the ethical standards of the institutional and national research committees and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study was approved by the Ethical Review Board of Kitasato University and its affiliated hospitals (approval number B21-095). The use of the opt-out method in lieu of written informed consent was approved by the ethics review boards of the participating hospitals.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

The authors declare that no funds, grants, or other supports were received during the preparation of this manuscript.

Author Contributions

Yuki Akazawa and Satoshi Igawa conceived the study, participated in its design and coordination, and performed statistical analyses and interpretation. Hiroya Manaka, Kaori Yamada, Yuri Yagami, Nobuki Kaizuka, Yuki Akazawa, Hiroki Yamamoto, and Masashi Kasajima performed data curation. Yoshiro Nakahara, Takashi Sato, Hisashi Mitsufuji, Masanori Yokoba, Masaru Kubota, Jiichiro Sasaki, and Katsuhiko Naoki supervised the study. Yuki Akazawa, Satoshi Igawa, and Katsuhiko Naoki drafted the manuscript. All authors read and approved the final manuscript.

Funding Statement

The authors declare that no funds, grants, or other supports were received during the preparation of this manuscript.

Data Availability Statement

Data are not publicly available due to ethical reasons. Further inquiries can be directed to the corresponding author.