Clinical benefit of COX-2 inhibitors in the adjuvant chemotherapy of advanced non-small cell lung cancer: A systematic review and meta-analysis

Abstract

BACKGROUND

Lung cancer is a major cause of death among patients, and non-small cell lung cancer (NSCLC) accounts for more than 80% of all lung cancers in many countries.

AIM

To evaluate the clinical benefit (CB) of COX-2 inhibitors in patients with advanced NSCLC using systematic review.

METHODS

We searched the six electronic databases up until December 9, 2019 for studies that examined the efficacy and safety of the addition of COX-2 inhibitors to chemotherapy for NSCLC. Overall survival (OS), progression free survival (PFS), 1-year survival rate (SR), overall response rate (ORR), CB, complete response (CR), partial response (PR), stable disease (SD), and toxicities were measured with more than one outcome as their endpoints. Fixed and random effects models were used to calculate risk estimates in a meta-analysis. Potential publication bias was calculated using Egger’s linear regression test. Data analysis was performed using R software.

RESULTS

The COX-2 inhibitors combined with chemotherapy were not found to be more effective than chemotherapy alone in OS, progression free survival, 1-year SR, CB, CR, and SD. However, there was a difference in overall response rate for patients with advanced NSCLC. In a subgroup analysis, significantly increased ORR results were found for celecoxib, rofecoxib, first-line treatment, and PR. For adverse events, the increase in COX-2 inhibitor was positively correlated with the increase in grade 3 and 4 toxicity of leukopenia, thrombocytopenia, and cardiovascular events.

CONCLUSION

COX-2 inhibitor combined with chemotherapy increased the total effective rate of advanced NSCLC with the possible increased risk of blood toxicity and cardiovascular events and had no effect on survival index.

Key Words: Non-small cell lung cancer, COX-2, Survival, Progression free survival, Systematic review, Randomized controlled trials

Core Tip: This study demonstrated that in patients who received adjuvant chemotherapy for advanced non-small cell lung cancer, COX-2 inhibitors improved the overall response rate and had no improvement on prolonged mortality. However, COX-2 enhanced both the overall response rate and the 1-year survival rate. Concerning toxicity, celecoxib plus chemotherapy resulted in a higher incidence of hematologic toxicities. Meanwhile, rofecoxib may augment the risk of cardiovascular events.

INTRODUCTION

The proportion of non-small cell lung cancer (NSCLC) is more than 80% of all lung tumors. Most patients have advanced NSCLC at stage ШB or IV when diagnosed and confirmed and have to receive alleviative treatment in order to maintain their lives[1,2]. The median survival time is 6-10 mo for patients who are diagnosed with advanced NSCLC in performance status 0-2 while adopting palliative first-line chemotherapy[3-5]. For decades, chemotherapy has been the cornerstone of standard cancer treatment[6]. At present, the efficacy of various chemotherapy regimens has reached its peak[7]. New treatment strategies are hypothesized to improve the clinical benefit (CB) of advanced NSCLC.

Increased COX-2 expression was reported in close to 70% of NSCLC adeno-carcinomas[8,9]. COX-2 expression is upregulated in the early stage of tumori-genesis, and it can lead to poor prognosis by promoting tumor cell proliferation, angiogenesis, invasion, and metastasis[10-12]. By any reasonable assessment, this targeted treatment initially achieved great success but also produced unpredictable and occasionally serious side effects. Comparison between nonselective nonsteroidal anti-inflammatory drugs and rofecoxib has shown that rofecoxib contributes to a decrease of gastrointestinal hemorrhage but not a decrease of thrombosis[13]. However, with respect to adverse events, celecoxib has no significant improvement on decreasing gastrointestinal events. The meta-analysis by Chen et al[14] reported that celecoxib has a positive influence on the treatment of advanced cancers but increased the risk of cardiovascular events by using celecoxib, which cannot be ignored. Other studies[15-17] indicated that celecoxib increased the overall response rate (ORR) of advanced NSCLC with no significant difference in cardiovascular events. The study related to COX-2 for intervention of NSCLC is mired in controversy in the medical field. Therefore, this systematic review based on randomized controlled trials (RCTs) was conducted to appraise the benefit of chemotherapy-assisted addition of COX-2 for advanced NSCLC.

MATERIALS AND METHODS

Search strategy

Six electronic databases, including the MEDLINE and EMBASE from Ovid, the Cochrane Library, CNKI, Wanfang Date, and CBD, were searched through December 9, 2019 using “cyclooxygenase-2 inhibitor,” “COX-2,” “apricoxib,” “celecoxib,” “rofecoxib,” “non-small cell lung cancer,” “NSCLC,” and “randomized controlled trial.”

Inclusion and exclusion criteria

The following inclusion criteria for clinical trials: (1) The language was limited to Chinese and English; (2) The benefit of the addition of COX-2 to chemotherapy (the principle of quantitative simplicity) were compared; (3) The NSCLC stage IIIB or IV patients used were defined and confirmed; (4) Outcomes such as overall survival (OS), progression free survival (PFS), 1-year survival rate (SR), ORR, CB, complete response (CR), partial response (PR), stable disease (SD), and toxicities were measured with more than one outcome as their endpoints. The primary outcomes were the OS, PFS, 1-year SR, ORR, and CB. The rate of CR and PR and the rate of grade 3 and 4 toxicity are regarded as the secondary endpoints; and (5) The study type was RCT.

Studies with criteria were excluded: (1) Patients experienced chemotherapy, immunotherapy, or any systemic therapy for NSCLC before; (2) The study was a duplicate; and (3) The data could not be extracted or obtained through contact with the author.

Data extraction and methodological quality

The data extracted were study design, patient characteristics, interventions, controls, and outcomes. The data acquisition was done independently by two authors.

The methodological quality was mainly focused on five aspects, including randomization methods, stratification factors, double blind, follow-up, and intent to treat, which were independently evaluated by two commentators. If there was a dispute, a third reviewer was consulted.

Statistical analysis

The hazard ratio (HR) was considered a reasonable effect size for OS and PFS outcomes after careful consideration. The existing HR with 95% confidence interval (CI) values provided from the original research, and then HR data was obtained. If the HR and 95%CI values were not provided, the Kaplan-Meier survival curve[18] was adopted. The relative risk (RR) with 95%CI was employed for other dichotomous outcomes[19,20]. The statistical test was performed for heterogeneity, and I² > 40% and P < 0.1 were considered as evidence for heterogeneity as well[20]. There is a theory that if the condition of the data were homogeneous under a fixed-effects model, then the heterogeneity of the results was derived from the type of Cox inhibitor and the difference in treatment line. Based on these modifiers, subgroups were conducted to address and analyze the heterogeneity. Ideally, the data are still heterogeneous, for which we introduce a stochastic effect model. The fixed-effects model was used instead when I² ≤ 40%[21]. Besides treatment line (first-line and second-line) and phase (II and III), COX-2 inhibitor types (celecoxib, rofecoxib, and apricoxib) were also identified as significant source of heterogeneity. Egger’s test was a methodological tool to solve quantitative detection publication bias[20]. All data analyses were performed by R 5.3.1 software.

RESULTS

Characteristics of included studies

There are 1328 publications picked from the six databases (Figure 1). Ultimately, 12 studies[22-33] involving 2273 patients were screened and included in this meta-analysis. The COX-2 inhibitors, including celecoxib, apricoxib, and rofecoxib, were adopted in these studies and with most of the trials opting for celecoxib. Only three studies chose rofecoxib or apricoxib. Table 1 showed the characteristics of the 12 studies.

Figure 1 Summary of trial identification and selection.

Table 1 Characteristics of each individual study.

Trials or Ref.	Year	Phase	Study period	Country	Sample (I/C)	Age (I/C)	Male (female) (I/C)	Histology (I/C) (AC, SCC, Other)	Extent of disease, Stage	ECOG PS or Karnofsky score	Treatment Line	Interventions	Control	Follow-up in mo
Lilenbaum et al[22]	2006	II	Feb 2002 to Sept 2003	United States	133 (67/66)	62.7 (37-84)/63.5 (41-78)	40 (27)/40 (26)	NA, NA, NA	ШB, IV	ECOG 0-1	Second	Celecoxib 400 mg po bid + DTX 35 mg/m2 or GEM 1000 mg/m2 + CPT-11 60-100 mg/m2 ivgtt day 1 and day 8, q3w	DTX 35 mg/m2 or GEM 1000 mg/m2 + CPT-11 60-100 mg/m2 ivgtt day 1 and day 8, q3w	NA
GECO[23]	2007	Ш	Jan 2003 to May 2005	Italy	400 (149/251)	61.5 (29-71)/59.0 (37-70)	120 (29)/202 (49)	68/134, 47/53, 34/64	ШB, IV	ECOG 0-1	First	Rofecoxib 50 mg po qd + GEM 1200 mg/m2 in 30-min or PCI GEM 1200 mg/m2 over 120-min iv infusions days 1 and 8 + DDP 80 mg/m2 ivgtt qd day 1, q3w	GEM 1200 mg/m2 in 30-min or PCI GEM 1200 mg/m2 over 120-min iv infusions days 1 and 8 + DDP 80 mg/m2 ivgtt qd day 1, q3w	22
Zhou et al[29]	2007	II	June 2004 to June 2005	China	65 (32/33)	57.0 (45-77)/55.5 (40-76)	24 (8)/24 (9)	17/19, 9/8, 5/3	ШB, IV	ECOG 0-2	First	Celecoxib 400 mg po bid days 1-12 + NVB 25 mg/m2 iv qd day 1 and 8 + DDP 75 mg/m2 ivgtt qd days 1 and 2, q3w	NVB 25 mg/m2 iv qd days 1 and 8 + DDP 75 mg/m2 ivgtt qd days 1 and 2, q3w	NA
Xiong et al[28]	2008	II	Jan 2003 to Jan 2006	China	60 (30/30)	56.4/58.3	16 (14)/17 (13)	16/17, 10/10, 4/3	ШB, IV	ECOG 0-2	First	Celecoxib 400 mg po bid + NVB 25 mg/m2 iv qd days 1 and 8 + DDP 70 mg/m2 ivgtt qd days 1-3, q3w	NVB 25 mg/m2 iv qd days 1 and 8 + DDP 70 mg/m2 ivgtt qd days 1-3, q3w	NA
CYCLUS[24]	2011	Ш	May 2003 to May 2006	Sweden	316 (158/158)	66 (38-85)/65 (37-85)	73 (85)/87 (71)	77/94, 38/27, 43/36	ШB, IV	ECOG 0-2	First	Celecoxib 400 mg po bid + GEM or NVB + CBP or DDP, ivgtt q3w1	Placebo + GEM or NVB + CBP or DDP, ivgtt q3w	36
NVALT-4[25]	2011	Ш	July 2003 to Dec 2007	Netherlands	561 (281/280)	62 (40-84)/61 (33-84)	184 (97)/171 (109)	138/132, 44/57, 99/91	ШB, IV	ECOG 0-2	First	Celecoxib 400 mg po bid + DTX 75 mg/m2 ivgtt qd day 1 + CBP ivgtt qd day 1, q3w2	Placebo + DTX 75 mg/m2 ivgtt qd day 1 + CBP ivgtt qd day 1, q3w	NA
Liu et al[30]	2012	NA	Jan 2006 to May 2011	China	46 (24/22)	62 (49-75)/64 (52-76)	14 (10)/15 (7)	15/14, 9/8, 0/0	ШB, IV	Karnofsky ≥ 70	First	Celecoxib 400 mg po bid days 1-5 + DTX 75 mg/m2 ivgtt qd day 1 + DDP 100 mg/m2 ivgtt qd day 1, q3w	DTX 75 mg/m2 ivgtt qd day 1 + DDP 100 mg/m2 ivgtt qd day 1, q3w	NA
Sörenson et al[32]	2013	Ш	May 2006 to May 2009	Sweden	107 (52/55)	65 (37-84)	50/57	65, 16, 26	ШB, IV	NA	First	Celecoxib at a dose of 400 mg bid + carboplatin plus gemcitabine/vinorelbine	Carboplatin + gemcitabine/ vinorelbine	5
Gitlitz et al[33]	2014	II	NA	United States	120 (78/42)	63 (35-81)/65 (36-84)	78 (42)/42 (25)	45/24, 21/11, 12/7	ШB, IV	ECOG 0-2	Second	Apricoxib (400 mg/d) + erlotinib (150 mg/d) on 21-d cycles	Placebo + erlotinib (150 mg/d) on 21-d cycles	NA
0822-GCC[26]	2015	II	NA	United States	72 (36/36)	62/66	20 (16)/20 (16)	24/25, 8/6, 4/5	ШB, IV	ECOG 0-2	Second	Apricoxib 400 mg po qd + DTX 75 mg/m2 or PET 500 mg/m2, q3w	Placebo 400 mg po qd DTX 75 mg/m2 or PET 500 mg/m2, q3w	NA
Teng et al[31]	2015	II	Aug 2009 to May 2012	China	81 (41/40)	57.7 (28-72)/57.3 (33-76)	30 (11)/26 (14)	28/26, 13/14, 0/0	ШB, IV	ECOG 0-1	First	Celecoxib 200 mg po bid + NVB 25 mg/m2 ivgtt days 1 and 8 + DDP 70 mg/m2 ivgtt qd day 1, q4w	NVB 25 mg/m2 ivgtt days 1 and 8 + DDP 70 mg/m2 ivgtt qd day 1, q4w	NA
CALGB-30801[27]	2017	Ш	Nov 2013 to Jan 2016	United States	312 (154/158)	64 (38-83)/64 (36-89)	82 (72)/87 (71)	NA, 44/43, NA	ШB, IV	ECOG 0-2	First	Celecoxib 400 mg po bid + CBP + PET 500 mg/m2 day 1, q3w for nonsquamous or Celecoxib 400 mg po bid + CBP day 1 + GEM 1000 mg/m2 day 1 and day 8, q3w for squamous	Placebo + CBP + PET 500 mg/m2 day 1, q3w for nonsquamous or placebo + CBP day 1 + GEM 1000 mg/m2 day 1 and day 8, q3w for squamous	31

¹The dose of chemotherapeutic agents was not mentioned in the trial. ²The dose of carboplatin was not mentioned in the trial. AC: Adenocarcinoma; I/C: Interventions/Control; Bid: Twice daily; CBP: Carboplatin; CR: Complete response; d: Day; DDP: Cisplatin; DTX: Docetaxel; iv: Intravenously; ECOG PS: Eastern Cooperative Oncology Group performance status; GEM: Gemcitabine; ivgtt: Intravenous drip; PCI: Prolonged constant infusion; NA: Not applicable; NVB: Vinorelbine; PET: Pemetrexed; OS: Overall survival; PFS: Progression free survival; Po: Orally; PR: Partial response; SCC: Squamous cell carcinoma; q: Every; w: Weeks.

Methodological quality

Of these 12 studies, only two trials[31,33] have not reported a random component in their sequence-generation process. Five studies[24,26,27,32,33] were designed with a double-blind trial. Although only five studies[9,12,23,24,27] described specific follow-up times, and all studies used intention-to-treat strategy in the evaluation of outcome measures with the exception of one study[29]. The result of methodological quality is shown in Table 2.

Table 2 The risk of bias in the included studies.

Trial or Ref.	Year	Randomization methods	Stratification factors	Double blind	Follow-up	Intent to treat
Lilenbaum et al[26]	2006	Centralized	ECOG PS, age, sex, disease stage, response to treatment	No	NA	Yes
GECO[23]	2007	Centralized	Treatment, gender, PS, disease stage, tumor histology, center (three categories according to size)	No	Median follow-up of 22 mo of alive patients (range 0-40)	Yes
Zhou et al[29]	2007	Envelopes	Types	No	NA	No: 4 of 65 excluded from analysis
Xiong et al[28]	2008	Random number table	Disease stage, COX-2 expression	No	NA	Yes
CYCLUS[24]	2011	Minimization	ECOG PS, sex, stage, smoking status	Yes	After randomization, the follow-up time ranged from 0 to 36 mo	Yes
NVALT-4[25]	2011	Centralized	PS, extent of disease, use of salicylic acid, histology, COX-2 expression, treatment	No	NA	Yes
Liu et al[30]	2012	Mechanical sampling method	Stage	No	NA	Yes
Sörenson et al[32]	2013	Minimization	ECOG PS, sex, stage, smoking status	Yes	After randomization, the follow-up time ranged from 0 to 36 mo	Yes
Gitlitz et al[33]	2014	NA	ECOG PS, sex, age	Yes	The median follow-up time was 30 mo	Yes
0822-GCC[26]	2015	Centralized	ECOG PS, sex, stage, race	Yes	NA	Yes
Teng et al[31]	2015	NA	Serum DKK-1 levels	No	NA	Yes
CALGB-30801[27]	2017	Stratified random permuted-blocks procedure	Sex, histology and chemotherapy, smoking status, stage, age group, PS	Yes	The median follow-up time was 31 mo	Yes

ECOG: Eastern Cooperative Oncology Group; PS: Performance status; COX-2: Cydooxygenase-2; NA: Not available.

Results of primary outcomes

OS: A total of seven studies showed that compared with chemotherapy alone the result of combinations of treatments revealed that there was no statistically significant difference in OS (HR = 1.08, 95%CI: 0.96 to 1.22; I²: 0%) (Figure 2).

Figure 2 Subgroup analyses of forest plot for overall survival.

In order to evaluate the CB of COX-2 inhibitors, the subgroup analyses were conducted according to the type of COX-2 inhibitor and treatment line. No CB in OS was observed among the groups: apricoxib (HR = 1.04, 95%CI: 0.64 to 1.69), celecoxib (HR = 1.10, 95%CI: 0.96 to 1.27), and rofecoxib (HR = 1.00, 95%CI: 0.75 to 1.34) (Figure 2A). Conducting subgroups by the type of treatment line compared with chemotherapy alone, the COX-2 inhibitors plus chemotherapy of first-line treatment (HR = 1.06, 95%CI: 0.93 to 1.21) and second-line treatment (HR = 1.19, 95%CI: 0.88 to 1.60) were not statistically different (Figure 2B). In subgroup analyses of phase, phase II (HR = 1.19, 95%CI: 0.88 to 1.60) and phase III (HR = 1.06, 95%CI: 0.93 to 1.21) were not remarkably different (Figure 2C).

PFS: Six RCTs involving 1794 patients presented the relative data for PFS. Compared with chemotherapy alone, the COX-2 inhibitors plus chemotherapy (Figure 3) also did not represent a significant difference in PFS (HR = 0.97, 95%CI: 0.86 to 1.10).

Figure 3 Subgroup analyses of forest plot for progression free survival.

Due to its lack of efficacy on PFS, we also performed further subgroup analysis, and all subgroup results were not significantly different (Figure 3A-C).

One-year SR: Eight RCTs including 1674 patients reported 1-year mortality rates (Figure 4). Compared with chemotherapy alone, the COX-2 inhibitors plus chemotherapy were not significantly different (RR = 1.11, 95%CI: 0.97 to 1.27).

Figure 4 Subgroup analyses of forest plot for 1-year survival rate.

Additionally, the results of the subgroup analysis were not significantly different among the types of COX-2 inhibitors: apricoxib (RR = 1.00, 95%CI: 0.15 to 6.72), celecoxib (RR = 1.12, 95%CI: 0.97 to 1.31), and rofecoxib (RR = 1.06, 95%CI: 0.78 to 1.44) (Figure 4A). However, when grouped by type of treatment line, the significant increase of 1-year SR (RR = 1.16; 95%CI: 1.01 to 1.34) was observed in first-line treatment, but there was no change in the second-line treatment (RR = 0.68; 95%CI: 0.41 to 1.14) (Figure 4B). In subgroup analyses of phase, phase II (HR = 1.14, 95%CI: 0.71 to 1.84) and phase III (HR = 1.08, 95%CI: 0.92 to 1.27) were not significantly different (Figure 4C).

ORR: Eight RCTs including 1662 patients reported ORRs. Comparison of two groups as shown in Figure 5 resulted in an increase in the ORR (RR = 1.28, 95%CI: 1.10 to 1.49).

Figure 5 Subgroup analyses of forest plot for overall response rate.

In the subgroup analysis, significantly increased ORRs were observed in celecoxib (RR = 1.23, 95%CI: 1.04 to 1.45), rofecoxib (RR = 1.56, 95%CI: 1.08 to 2.25), first-line treatment (RR = 1.30, 95%CI: 1.11 to 1.51), and phase III (RR = 1.27, 95%CI: 1.07 to 1.50). Second-line treatment (RR = 0.49, 95%CI: 0.09 to 2.60) and phase II (RR = 1.31, 95%CI: 0.88 to 1.95) with COX-2 inhibitors reported no significant differences (Figure 5A-C).

CB: Nine RCTs including 1776 patients reported a CB (Figure 6). Compared with chemotherapy alone, the COX-2 inhibitors plus chemotherapy did not represent a significant difference in CB (RR = 1.05, 95%CI: 0.98 to 1.11; I²: 0%).

Figure 6 Subgroup analyses of forest plot for clinical benefit.

As mentioned above, no significantly different results were found in the three subgroup analyses: apricoxib (RR = 1.10, 95%CI: 0.73 to 1.65; I²: NA), celecoxib (RR = 1.05, 95%CI: 0.99 to 1.12; I²: 14.4%), rofecoxib (RR = 0.99, 95%CI: 0.81 to 1.21; I²: NA), first-line treatment (RR = 1.05, 95%CI: 0.99 to 1.12; I²: 5.8%), second-line treatment (RR = 0.96, 95%CI: 0.69 to 1.33; I²: 0.0%), phase II (RR = 1.07, 95%CI: 0.92 to 1.26; I²: 0%), and phase III (RR = 1.53, 95%CI: 1.00 to 2.33; I²: NA) (Figure 6A-C).

Results of secondary outcome variables

CR: When we assessed the effect on CR involving eight RCTs (1460 patients, there were no differences between combined treatment and chemotherapy alone (RR = 0.90, 95%CI: 0.31-2.57) (Figure 7).

Figure 7 Subgroup analyses of forest plot for complete response.

The results of two subgroup analyses showed no significant difference: apricoxib (RR = 0.17, 95%CI: 0.01 to 4.18), celecoxib (RR = 0.75, 95%CI: 0.19 to 3.05), rofecoxib (RR = 5.08, 95%CI: 0.25 to 104.78), first-line treatment (RR = 1.18, 95%CI: 0.36 to 3.85), second-line treatment (RR = 0.17, 95%CI: 0.01 to 4.18), phase II (RR = 0.74, 95%CI: 0.12 to 4.38), and phase III (RR = 0.84, 95%CI: 0.03 to 28.0) (Figure 7A-C).

PR: When we assessed the effect on PR involving eight RCTs (1460 patients, COX-2 inhibitors combined with chemotherapy had a significant increase (RR = 1.31, 95%CI: 1.11 to 1.56) compared with chemotherapy alone (Figure 8).

Figure 8 Subgroup analyses of forest plot for partial response.

The following details of subgroup analysis were represented, and the significantly increased ORRs were observed for celecoxib (RR = 1.27, 95%CI: 1.04 to 1.55), rofecoxib (RR = 1.49, 95%CI: 1.03 to 2.16), first-line treatment (RR = 1.34, 95%CI: 1.13 to 1.60), and phase III (RR = 1.33, 95%CI: 1.09 to 1.63). Apricoxib (RR = 1.17, 95%CI: 0.38 to 3.56), second-line treatment (RR = 0.88, 95%CI: 0.35 to 2.17), and phase II (RR = 1.26, 95%CI: 0.86 to 1.84) with COX-2 inhibitors showed no remarkably differences (Figure 8A-C).

SD: When we assessed the effect on SD involving nine RCTs with 1776 patients, COX-2 inhibitors plus chemotherapy resulted in a significant increase in SD (RR = 0.90, 95%CI: 0.80 to 1.02) compared with chemotherapy alone (Figure 9).

Figure 9 Subgroup analyses of forest plot for stable disease.

Subgroup analysis showed an insignificant increase in SD for apricoxib (RR = 1.16, 95%CI: 0.68 to 1.97), celecoxib (RR = 0.94, 95%CI: 0.83 to 1.07), first-line treatment (RR = 0.89, 95%CI: 0.79 to 1.01), second-line treatment (RR = 1.02, 95%CI: 0.68 to 1.52), phase II (RR = 0.99, 95%CI: 0.78 to 1.27), and phase III (RR = 0.84, 95%CI: 0.66 to 1.07). However, a change was noted for rofecoxib (RR = 0.57, 95%CI: 0.37 to 0.87).

Toxicity: The increase in COX-2 inhibitor was positively correlated with the increase in grade 3 and 4 toxicity of leukopenia (RR = 1.20, 95%CI: 1.03 to 1.40), thrombocytopenia (RR = 1.33, 95%CI: 1.05 to 1.68), and cardiovascular events (RR = 2.39, 95%CI: 1.06 to 5.42) (Table 3).

Table 3 Subgroup analyses of the toxicities of COX-2 inhibitor.

Toxicity	RCT, n	RR (95%CI)	P value for between groups	Toxicity	RCT, n	RR (95%CI)	P value for between groups
Leucopenia	8	1.20 (1.03, 1.40)	0.020	Diarrhea	3	1.31 (0.64, 2.71)	0.460
COX-2 inhibitor type				COX-2 inhibitor type
Celecoxib	6	1.26 (1.07, 1.49)	0.280	Celecoxib	2	1.24 (0.59, 2.62)	0.940
Rofecoxib	1	0.80 (0.43, 1.50)		Rofecoxib	1	3.05 (0.13, 74.1)
Apricoxib	1	0.92 (0.47, 1.80)		Apricoxib	1	2.69 (0.33, 22.3)
Treatment line				Treatment line
First-line	6	1.20 (1.02, 1.42)	0.900	First-line	2	0.91 (0.40, 2.07)	0.080
Second-line	2	1.19 (0.76, 1.87)		Second-line	2	4.10 (0.95, 17.60)
Phase				Phase
II	4	1.14 (0.77, 1.69)	0.720	II	2	4.10 (0.95, 17.60)	0.080
III	4	1.21 (1.03, 1.44)		III	2	0.91 (0.40, 2.07)
Thrombocytopenia	8	1.33 (1.05, 1.68)	0.017	Gastric ulcer	2	1.00 (0.25, 3.97)	0.997
COX-2 inhibitor type				COX-2 inhibitor type
Celecoxib	6	1.40 (1.08, 1.81)	0.560	Celecoxib	2	1.00 (0.25, 3.97)	NA
Rofecoxib	1	1.02 (0.59, 1.76)		Rofecoxib	NA	NA
Apricoxib	1	3.00 (0.13, 71.30)		Apricoxib	NA	NA
Treatment line				Treatment line
First-line	6	1.24 (0.97, 1.58)	0.090	First-line	2	1.00 (0.25, 3.97)	NA
Second-line	2	2.66 (1.14, 6.17)		Second-line	NA	NA
Phase				Phase
II	4	2.69 (1.19, 6.07)	0.070	II	2	1.00 (0.25, 3.97)	NA
III	4	1.23 (0.96, 1.56)		III	NA	NA
Anemia	5	1.32 (0.75, 2.33)	0.343	Asthenia	7	0.84 (0.56, 1.28)	0.426
COX-2 inhibitor type				COX-2 inhibitor type
Celecoxib	3	2.76 (0.96, 7.97)	0.110	Celecoxib	5	0.94 (0.60, 1.48)	0.590
Rofecoxib	1	0.80 (0.38, 1.69)		Rofecoxib	1	0.51 (0.16, 1.64)
Apricoxib	2	3.14 (0.51, 19.50)		Apricoxib	2	0.94 (0.20, 4.44)
Treatment line				Treatment line
First-line	3	1.07 (0.56, 2.05)	0.140	First-line	5	0.92 (0.60, 1.42)	0.560
Second-line	3	2.91 (0.89, 9.98)		Second-line	3	0.53 (0.15, 1.88)
Phase				Phase
II	4	3.03 (1.00, 9.24)	0.100	II	4	0.75 (0.28, 2.02)	0.900
III	2	1.01 (0.52, 1.97)		III	3	0.86 (0.54, 1.39)
Nausea	7	0.85 (0.53, 1.36)	0.507	Cardiotoxicity	5	2.39 (1.06, 5.42)	0.037
COX-2 inhibitor type				COX-2 inhibitor type
Celecoxib	5	0.87 (0.50, 1.51)	0.960	Celecoxib	3	1.55 (0.53, 4.50)	0.540
Rofecoxib	1	0.76 (0.27, 2.13)		Rofecoxib	1	4.58 (1.01, 20.70)
Apricoxib	2	1.00 (0.15, 6.72)		Apricoxib	1	3.00 (0.13, 71.30)
Treatment line				Treatment line
First-line	6	0.84 (0.52, 1.37)	0.860	First-line	4	2.35 (1.01, 5.49)	0.880
Second-line	2	1.00 (0.15, 6.72)		Second-line	1	3.00 (0.13, 71.30)
Phase				Phase
II	4	1.44 (0.58, 3.59)	0.400	II	1	3.00 (0.13, 71.30)	0.880
III	3	0.67 (0.36, 1.25)		III	4	2.35 (1.01, 5.49)
Neurotoxicity	4	1.02 (0.23, 4.45)	0.977
COX-2 inhibitor type
Celecoxib	3	1.02 (0.18, 5.83)	0.100
Rofecoxib	1	1.02 (0.06, 16.07)
Apricoxib	NA	NA
Treatment line
First-line	4	1.02 (0.23, 4.45)	1.000
Second-line	NA	NA
Phase
II	2	3.09 (0.13, 73.20)	0.420
III	2	0.68 (0.11, 4.04)

RCT: Randomized controlled trial; RR: Relative risk; CI: Confidence interval; NA: Not applicable.

Subgroup analysis of leukopenia in Table 3 showed that all of celecoxib (RR = 1.26, 95%CI: 1.07 to 1.49), first-line treatment (RR = 1.20, 95%CI: 1.02 to 1.42), and phase III (RR = 1.21, 95%CI: 1.03 to 1.44) increased the risk of leukopenia. Subgroup analysis of thrombocytopenia showed that celecoxib (RR = 1.40, 95%CI: 1.08 to 1.81), second-line treatment (RR = 2.66, 95%CI: 1.14 to 6.17), and phase II (RR = 2.69, 95%CI: 1.19 to 6.07) significantly increased the incidence of thrombocytopenia. Subgroup analysis of cardiovascular events showed that rofecoxib (RR = 4.58, 95%CI: 1.01 to 20.7), first-line treatment (RR = 2.35, 95%CI: 1.01 to 5.49), and phase III (RR = 2.35, 95%CI: 1.01 to 5.49) increased the risk of cardiovascular events. However, the risks of other toxicities were not found to be increased significantly (Table 3).

Publication bias: In the results of publication bias using Egger’s test, all primary outcomes (P_OS: 0.314, P_PFS: 0.807, P_ORR: 0.883, P_{1-year SR}: 0.624, and P_CB: 0.220) were not significantly different. With respect to secondary outcomes, we did not obtain significant difference (data not shown).

DISCUSSION

Based on extensive preclinical and clinical studies, COX-2 inhibitors have shown significant CBs in both therapy and the chemoprevention of lung cancer. In this study, COX-2 inhibitors can increase the efficacy of chemotherapy regarding ORR. In a subgroup analysis, we found that celecoxib and rofecoxib might improve the ORR of patients with advanced NSCLC. Based on the treatment line, an increased ORR was found in first-line treatment with COX-2 inhibitors for advanced NSCLC patients. However, the second-line treatment with COX-2 inhibitors did not yield a significant effect in the ORR, possibly due to the inclusion of only one article. Teng et al[31] reported a higher ORR with celecoxib added to chemotherapy, whereas a study by Schneider et al[34] showed that celecoxib did not seem to improve the response rate. The most plausible explanation may be that different chemotherapy regimens were used. Teng et al[31] used gemcitabine/cisplatin, whereas Schneider et al[34] used docetaxel. However, for the CB, a significant difference was not discovered. The findings of the subgroup analysis were consistent with those of previous studies[22,28,30]. Although no evidence showed that COX-2 inhibition could improve the CB for advanced NSCLC patients, Edelman et al[35] highlighted the importance of seeking molecular oriented therapy using COX-2 inhibitors. COX-2 inhibitors plus chemotherapy has no improvement on the 1-year SR for advanced NSCLC patients. In the subgroup analysis of the treatment line, COX-2 inhibitors in first-line treatment revealed a significant increase of the 1-year SR. Accordingly, the conclusion was made that COX-2 inhibitors more effectively improved both ORR and the 1-year SR for people suffering with advanced NSCLC using first-line chemotherapy. The meta-analysis by Zhou et al[16] stated that the COX-2 inhibitors may increase the ORR with advanced NSCLC.

Toxicity exists differently for individuals in incidence and severity[36]. Compared to chemotherapy alone, COX-2 inhibitors associated with chemotherapy might have a higher incidence of hematological toxicity, except for anemia. In addition, it was confirmed by subgroup analysis that combined treatment (celecoxib plus chemotherapy) could increase the risk of hematological toxicity, particularly for two periods (in that first-line treatment with leukopenia and second-line treatment with thrombocytopenia). This is consistent with previous meta-analyses[15,16]. Nevertheless, it is likely that COX-2 is necessary for marrow recovery after cytotoxic chemo-therapy[37]. A study[38] suggested that the directed differentiation of erythroid, myeloid, and megakaryocytic progenitors is related to the level of COX-2. Therefore, COX-2 inhibitors may also result in higher risk of hematological toxicity while increasing the ORR by using COX-2 inhibitors.

This study illustrated that COX-2 augmented the risk of cardiovascular events as well. Cardiovascular events with higher incidences happened when using rofecoxib. The influence of rofecoxib on cardiovascular events still needed to be investigated for a few studies, whereas celecoxib had no effect on cardiovascular events. On the basis of classifying the treatment line, it slightly increased the risk of cardiovascular events of advanced NSCLC through first-line treatment associated with COX-2, but no obvious differences were observed for second-line treatment. Prostacyclin[39], a substance that associates with the expression of COX-2, existed in rofecoxib. Therefore, rofecoxib might participate in the process of formation of thrombosis. In vitro experiments have proven that celecoxib has a lower specific effect on COX-2 than rofecoxib and is less likely to cause thrombosis, which indicates the rationality of our hypothesis. Given that patients may not benefit from COX-2 selective nonsteroidal anti-inflammatory drugs[40], it makes sense to use aspirin to prevent vascular events again.

There are several meta-analyses concerning published research on the CB profile of COX-2[14-17]. The superiority of the ORR alone made it difficult to adequately demonstrate that the inhibition of the COX-2 inhibitors could improve the efficacy. A relevant study[15] analyzed six studies, setting forth all endpoints that did not conduct subgroup analysis. In addition, no subgroup analyses were performed when toxicities were assessed by Zhou et al[16]. Dai et al[17] study describing all efficacy endpoints with subgroup analysis, but other efficacy outcomes (CB, CR, PR and SD) were lacking, and toxicity was not performed by subgroup analysis to explore the difference in different types of COX-2 inhibitors and the treatment line. In this meta-analysis, 12 studies were included, and five main outcomes (ORR, CB, 1-year SR, OS and PFS) and four secondary outcomes (CR, PR, SD and toxicity) were defined above. Moreover, considering the potential clinical heterogeneity, subgroup analyses were employed based on the different types of COX-2 inhibitors and treatment line.

This study has some limitations. First, there are not many clinical trials that met the study design of this systematic review, especially in subgroup analysis, the small number of trials for rofecoxib, apricoxib, or second-line treatment limited the analytical power. Hence, more clinical studies are needed to further confirm our results about combined treatment and chemotherapy alone for advanced NSCLC. Second, due to the lack of data on the response rate and survival outcomes in the included RCTs, this may result in too small a result sample and the accuracy of the results.

CONCLUSION

This meta-analysis demonstrated that, in terms of ORR for patients who received adjuvant chemotherapy of advanced NSCLC, COX-2 inhibitors improved the ORR and have no improvement on prolonged mortality. However, the COX-2 inhibitors could enhance both the ORR and improve the 1-year SR, particularly with first-line chemotherapy. Concerning toxicity, celecoxib plus chemotherapy resulted in a higher incidence of hematologic toxicities. Meanwhile, rofecoxib may augment the risk of cardiovascular events.

ARTICLE HIGHLIGHTS

Research background

The proportion of non-small cell lung cancer (NSCLC) is more than 80% of all lung tumors. Most patients have advanced NSCLC at stage ШB or IV when diagnosed and have to receive alleviative treatment in order to maintain their lives. The median survival time is 6-10 mo for patients who are diagnosed with advanced NSCLC in performance status 0-2 when adopting palliative first-line chemotherapy.

Research motivation

The motivation of this study is to investigate COX-2 for intervention of NSCLC, which is mired in controversy in the medical field.

Research objectives

This systematic review based on randomized controlled trials was conducted to appraise the benefit of chemotherapy-assisted addition of COX-2 for advanced NSCLC.

Research methods

We searched the six electronic databases up until December 9, 2019 for studies that examined the efficacy and safety of the addition of COX-2 inhibitors to chemotherapy for NSCLC. Overall survival(OS), progression free survival (PFS), 1-year survival rate (SR), overall response rate (ORR), clinical benefit (CB), complete response (CR), partial response (PR), stable disease (SD), and toxicities were measured with more than one outcome as their endpoints. Fixed and random effects models were used to calculate risk estimates in a meta-analysis. Potential publication bias was calculated using Egger’s linear regression test. Data analysis was performed using R software.

Research results

The COX-2 inhibitors combined with chemotherapy were not found to be more effective than chemotherapy alone in OS, PFS, 1-year SR, CB, CR, and SD. However, there was a difference in ORR for patients with advanced NSCLC. In a subgroup analysis, significantly increased ORR results were found for celecoxib, rofecoxib, first-line treatment, and PR. For adverse events, the increase in COX-2 inhibitor was positively correlated with the increase in grade 3 and 4 toxicity of leukopenia, thrombocytopenia and cardiovascular events.

Research conclusions

COX-2 inhibitor combined with chemotherapy increased total effective rate of advanced NSCLC with the possible increased risk of blood toxicity and cardiovascular events and had no effect on survival index.

Research perspectives

This study can provide reference value for the application of COX-2 in the treatment of lung cancer.