Phase III trial of gemcitabine and carboplatin versus mitomycin, ifosfamide, and cisplatin or mitomycin, vinblastine, and cisplatin in patients with advanced nonsmall cell lung carcinoma

Abstract

BACKGROUND

The authors compared gemcitabine and carboplatin (GC) with mitomycin, ifosfamide, and cisplatin (MIC) or mitomycin, vinblastine, and cisplatin (MVP) in patients with advanced nonsmall cell lung carcinoma (NSCLC). The primary objective was survival. Secondary objectives were time to disease progression, response rates, evaluation of toxicity, disease-related symptoms, World Health Organization performance status (PS), and quality of life (QoL).

METHODS

Three hundred seventy-two chemotherapy-naïve patients with International Staging System Stage III/IV NSCLC who were ineligible for curative radiotherapy or surgery were randomized to receive either 4 cycles of gemcitabine (1000 mg/m² on Days 1, 8, and 15) plus carboplatin (area under the serum concentration-time curve, 5; given on Day 1) every 4 weeks (the GC arm) or MIC/MVP every 3 weeks (the MIC/MVP arm).

RESULTS

There was no significant difference in median survival (248 days in the MIC/MVP arm vs. 236 days in the GC arm) or time to progression (225 days in the MIC/MVP arm vs. 218 days in the GC arm) between the 2 treatment arms. The 2-year survival rate was 11.8% in the MIC/MVP arm and 6.9% in the GC arm. The 1-year survival rate was 32.5% in the MIC/MVP arm and 33.2% in the GC arm. In the MIC/MVP arm, 33% of patients responded (4 complete responses [CRs] and 57 partial responses [PRs]) whereas in the GC arm, 30% of patients responded (3 CRs and 54 PRs). Nonhematologic toxicity was comparable for patients with Grade 3–4 symptoms, except there was more alopecia among patients in the MIC/MVP arm. GC appeared to produce more hematologic toxicity and necessitated more transfusions. There was no difference in performance status, disease-related symptoms, or QoL between patients in the two treatment arms. Fewer inpatient stays for complications were required with GC.

CONCLUSIONS

The results of the current study failed to demonstrate any difference in efficacy between the newer regimen of GC and the older regimens of MIC and MVP. Cancer 2003;98:542–53. © 2003 American Cancer Society.

DOI 10.1002/cncr.11535

Cytotoxic chemotherapy has an established role in the palliative management of patients with advanced nonsmall cell lung carcinoma (NSCLC). A meta-analysis of 11 randomized clinical trials comparing cisplatin-based chemotherapy with best supportive care alone showed that chemotherapy was associated with a 27% reduction in risk of death, a 10% improvement in 1-year survival, and an increased median survival of 1.5 months.1 Symptom benefit is described in 60–70% of patients, and there is an improvement in performance status (PS) in 30–40% of patients.2–4 Furthermore, chemotherapy improves quality of life (QoL) and is cost effective.2–5 Nevertheless, any survival benefit must be balanced against treatment-related toxicity. The cisplatin-based mitomycin, ifosfamide, and cisplatin (MIC) and mitomycin, vinblastine, and cisplatin (MVP) are standard regimens that often are used to treat patients with advanced NSCLC in the U.K. and Europe.6, 7 The platinum dose used in the U.K. (50–60 mg/m²) is lower than the 100–120 mg/m² dose favored elsewhere. High doses are difficult to sustain in this frequently comorbid population, so dose reductions often are required.8, 9 This may be ameliorated by modifying the dose of cisplatin to 50 mg/m², which produces a treatment that can be tolerated by our NSCLC population, many of whom have a PS of 2 at the initiation of therapy. These regimens are associated with low toxicity and have the potential to improve QoL.6, 10

Several novel chemotherapeutic agents currently are under evaluation for the treatment of patients with advanced NSCLC, including gemcitabine, vinorelbine, the taxanes, and the topoisomerase I inhibitors. Interest has focused on the use of gemcitabine, a deoxycytidine analogue with a favorable side-effect profile. An analysis of 4 Phase II trials with greater than 300 NSCLC patients demonstrated a single-agent response rate of 20% with a median survival of 8.1–9.2 months.4 Disease-related symptoms improved in treated patients, including pain (32%), cough (44%), dyspnea (26%), hemoptysis (63%), and anorexia (29%). The duration of improvement lasted 2–5 months. Gemcitabine acts synergistically with platinum chemotherapy and is a candidate for combination chemotherapy.11 A number of reports have investigated gemcitabine and cisplatin (GP) combinations, including a Phase III trial of 307 patients that compared GP and MIC12 with gemcitabine at a dose of 1000 mg/m² administered on Days 1, 8, and 15 and cisplatin at a dose of 100 mg/m² administered on Day 2. The GP regimen had a significantly higher response rate of 38% compared with 26% for MIC. However, there were no significant differences in QoL, time to progression, and overall survival. GP was found to be associated with more thrombocytopenia compared with MIC. However, this may be contributed to by the high cisplatin dose (100 mg/m²) used in GP, and the MIC myelotoxicity may have been under-estimated, because fewer blood samples were taken.

At the time that this trial was conceived, there was limited efficacy data regarding how carboplatin and cisplatin compared. Cisplatin or carboplatin combinations with etoposide had equivalent survival,13 whereas superior survival has been reported with carboplatin.14, 15 Recently, Rosell et al.16 compared paclitaxel plus carboplatin with paclitaxel plus cisplatin. Those two regimens yielded similar response rates, but significantly longer median survival was obtained with paclitaxel plus cisplatin. This conflicts with the data obtained from the recent, large Eastern Cooperative Oncology Group (ECOG) trial,17 which favored the paclitaxel plus carboplatin regimen, because there was equivalent efficacy but less toxicity with carboplatin compared with cisplatin. To our knowledge there has not yet been a direct comparison of GP versus gemcitabine plus carboplatin (GC). We chose carboplatin for this study because of the perception that it causes less toxicity compared with cisplatin.

The combination of gemcitabine plus carboplatin has been evaluated in Phase I and II trials.18–23 Gemcitabine given weekly at a dose of 1000 mg/m² for 3 of 4 weeks with carboplatin on Day 1 at a recommended dose of area under the serum concentration-time curve (AUC) 5. The dose-limiting toxicity was myelosuppression with a 31% response rate and a median survival of 45 weeks.18 The GC combination was considered active and was an attractive candidate for comparison with standard therapies in Phase III trials.

MATERIALS AND METHODS

The study protocol was a randomized Phase III trial with the patients enrolled in three centers.

Patient Selection

Eligible patients had to fulfill the following criteria: pathologically confirmed NSCLC; Stage III–IV disease;24 ineligible for curative radical radiotherapy or surgery after discussion in a multidisciplinary team setting comprised of at least a chest physician, a surgeon, and an oncologist;25 no previous chemotherapy; a PS of 0–2 (patients with a PS of 3 could be entered at the investigator's discretion); age older than 18 years; life expectancy of at least 12 weeks; adequate bone marrow reserve (leukocyte count > 3 × 10⁹/L, platelet count > 100 × 10⁹/L, and hemoglobin > 100 g/dL); creatinine clearance > 60 mL/minute; and adequate birth-control measures.

Patients were excluded if they had an active infection, bony disease as the only measurable disease, prior radiotherapy to the sole site of assessable disease, inadequate renal or hepatic function, serious comorbidity, other malignancy (except in situ carcinoma of the cervix or adequately treated basal cell carcinoma of the skin), peripheral neuropathy Grade > 2 (according to the National Cancer Institute Common Toxicity Criteria), significant neurologic or psychiatric disorder, or symptomatic brain metastases. The tumor had to be measurable unidimensionally or bidimensionally. All patients provided written, informed consent prior to randomization. Baseline evaluation included a complete history and examination, including measurement of palpable or visible disease, full blood count and biochemical profile, appropriate radiology, symptom scoring, PS, and European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 questionnaire, and Hospital Anxiety and Depression Scores (HADS) QoL questionnaire.

Treatment Schedule

The patients were randomized, using permuted blocks, by the Medical Statistics Department at Christie Hospital between GC and MIC or MVP (depending on the center preference) and were stratified by center, disease stage, and PS. On the GC arm, gemcitabine was given at a dose of 1000 mg/m² in 500 mL 0.9% saline intravenously (i.v.) over 30 minutes on Days 1, 8, and 15. Carboplatin dosed at AUC 5 (calculated according to the Calvert formula) i.v.26 was given on Day 1 only. The cycle was repeated every 28 days, if there was hematologic recovery, to a maximum of 4 cycles. For Day 1, chemotherapy was delayed by 1 week if the total leukocyte count was < 3 × 10⁹/L, neutrophil count was < 1.5 × 10⁹/L, or platelet count was < 100 × 10⁹/L. If there were three consecutive delays, then chemotherapy was discontinued. For Days 8 and 15, the gemcitabine dose was adjusted as follows: for patients with neutrophils 0.5–0.99 × 10⁹/L and/or platelets 50–74 × 10⁹/L, there was a 50% reduction; and, for patients with neutrophils < 0.5 × 10⁹/L and/or platelets < 50 × 10⁹/L, the dose was omitted.

MIC was administered as mitomycin C at a dose of 6 mg/m² by i.v. bolus, ifosfamide at a dose of 3 g/m² with sodium mercaptoethanesulfonate by i.v. infusion, and cisplatin at a dose of 50 mg/m² by i.v. infusion.2 MVP was comprised of mitomycin C at a dose of 8 mg/m² by i.v. bolus, vinblastine at a dose of 6 mg/m² (maximum 10 mg per dose) by i.v. bolus, and cisplatin at a dose of 50 mg/m² by i.v. infusion, with mitomycin omitted on Cycle 3. Both regimens were administered every 21 days, if there was hematologic recovery (the same parameters described above for the GC arm), to a maximum of 4 cycles. To proceed with each cycle, the creatinine clearance had to be > 50 mL per minute. There were no mandated dose reductions. The trial protocol suggested an antiemetic regimen comprised of metoclopramide, ondansetron, and dexamethasone that could be altered according to local practice and patient needs.

Objectives

The primary objective was difference in survival. The study was designed to have 80% power to detect a 10% survival advantage at 2 years, which required enrolling 160 patients into each arm. This number was increased with 10% uplift to allow for follow-up difficulties and loss of QoL data. Secondary objectives were time to disease progression, response rates, toxicity (including emergency hospitalization data, symptom control, and toxicity management), disease-related symptoms, PS, and QoL.

Treatment Evaluations

Treatment was evaluated immediately before each cycle and at 4 weeks and 8 weeks posttreatment, and the following parameters were documented: toxicities (National Cancer Institute Common Toxicity Criteria, version 1.0) and disease symptoms, using a descriptive scoring system with 0 indicating no symptoms, 1 indicating mild symptoms, 2 indicating moderate symptoms, and 3 indicating severe symptoms by physician assessment.4, 6, 27 The PS (World Health Organization [WHO]) and QoL also were recorded using the EORTC QLQ-C30 scale and the HADS depression scale.27–29 The EORTC QLQ-C30 scale29 is comprised of nine scales: one global scale, five functional scales (physical, role, emotional, cognitive, and social), three symptom scales (fatigue, pain, and nausea), and six single-item scales (dyspnea, sleep disturbance, appetite loss, constipation, diarrhea, and financial impact). The HADS score measures anxiety and depression. A score > 11 is considered significant.28

Any clinically evaluable disease was measured at baseline and was assessed after Cycle 2 and at the end of chemotherapy. Baseline and postchemotherapy computed tomography scans were performed; a chest radiograph was taken at baseline, prior to each course of chemotherapy, and after the completion of treatment. Response was evaluated radiologically according to the standard WHO criteria.30

Statistical Tests

Overall survival was evaluated by descriptive analysis using Kaplan–Meier estimates of the survival curves and the log-rank test for comparisons of the survival and time to progression curves. The patient characteristics, response rates, PS, and hematologic toxicities were compared using Pearson chi-square tests. Mann–Whitney U tests were used for transfusion requirements and hospitalization days, and frequency tables were used for changes in disease-related symptoms and QoL. QoL was presented as the percentage change from baseline to midcycle and from baseline to the end of treatment for the variables that had > 10% difference between the two treatment arms.28 The HADS scale was presented as the proportion of patients with scores < 11 or > 11 from baseline to midtreatment and from baseline to 4 weeks after administration of the last chemotherapy course.

RESULTS

Three hundred seventy-two patients with advanced NSCLC were recruited between January 8, 1997 and July 6, 2001. One hundred eighty-six patients were enrolled onto the GC arm, and 186 patients were enrolled onto the MIC/MVP arm, with 175 patients receiving MIC and 11 patients receiving MVP. Four patients did not receive their chosen treatment: One patient drew GC but refused it and, instead, received MIC off study; one patient had Day 1 GC but did not continue; one patient drew MIC but then withdrew from the study; and one patient was enrolled using incorrect data to assess eligibility and, in fact, was ineligible (Fig. 1).

Figure 1. Flow diagram of the clinical trial comparing gemcitabine and carboplatin (GC) with mitomycin, ifosfamide, and cisplatin (MIC) or mitomycin, vinblastine, and cisplatin (MVP; MIC/MVP) in the treatment of patients with nonsmall cell lung carcinoma.

Patient and Tumor Characteristics

Patient and tumor characteristics are listed in Table 1. The patient characteristics were balanced evenly for age and gender. The majority of patients were symptomatic at presentation with a PS of 1, 2, or 3 (81.0% of patients in the GC arm and 87.5% of patients in the MIC/MVP arm). The tumor characteristics are listed in Table 1 and were balanced evenly. The proportion of patients with Stage IIIA bulky disease (disease difficult to encompass in a radiotherapy field) in each arm was small, including only 11% of patients in the GC arm and 9% of patients in the MIC/MVP arm. The most common histologic type was squamous cell. Four percent of patients in the GC arm and 6% of patients in the MIC/MVP arm had undergone previous surgery.

Table 1. Patient Characteristics

Characteristic	GC		MIC/MVP
	No.	%	No.	%
GC: gemcitabine and carboplatin; MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin; WHO: World Health Organization; NOS, not otherwise specified.
No. of patients entered	186	—	186	—
Age (yrs)
Median	56	—	58	—
Range	23–85	—	36–79	—
Gender
Male	113	61	121	65
Female	73	39	65	35
WHO performance status
0	34	18	23	12
1	88	48	105	57
2	56	30	56	30
3	8	4	2	1
Stage
IIIA	20	11	17	9
IIIB	73	39	82	44
IV	93	50	87	47
Histology
Aquamous	91	49	85	46
Adenocarcinoma	52	28	61	33
Large cell	6	3	7	4
Undifferentiated	10	6	7	4
NOS	26	14	25	13
Prior surgery	8	4	12	6

Survival

At the time of analysis, 39 of 186 patients (21%) in the GC arm and 46 of 186 patients (25%) patients in the MIC/MVP arm remained alive. There was no statistical difference in overall survival between the two arms (236 days in the GC arm vs. 248 days in the MIC/MVP arm; P = 0.181). The 1-year survival rate was 33.2% (95% confidence interval [95% CI], 26–41%) in the GC arm and 32.5% (95% CI, 25–40%) in the MIC/MVP arm. The 2-year survival rate was 6.9% (95% CI, 3–13%) in the GC arm and 11.8% (95% CI, 6–19%) in the MIC/MVP arm (Fig. 2). Patients who had a PS of 2–3 had significantly worse survival compared with patients who had a PS of 0–1, with a median reported survival of 179 days and 282 days, respectively (P < 0.005) (data not presented).

Figure 2. Survival by treatment arm. MIC: mitomycin, ifosfamide, and cisplatin; or MVP: mitomycin, vinblastine, and cisplatin; Gem+Carbo: gemcitabine and carboplatin.

There were two treatment-related deaths. One patient had a cerebral hemorrhage 1 week after Cycle 3 of MIC chemotherapy during an episode of thrombocytopenia. The second patient, who also was receiving MIC, died 6 days after the initiation of Cycle 4 of pulmonary edema.

Time to Disease Progression

No statistically significant differences were recorded in the median time to disease progression (218 days in the GC arm vs. 225 days in the MIC/MVP arm). The 1-year progression free survival rate was 26% for patients in the MIC/MVP arm and 28% for patients in the GC arm.

Response Rates

Data for response to therapy were available for 361 patients, because some patients decided to terminate treatment or were not well enough for evaluation, and are shown in Table 2. In the MIC/MVP arm, 33% of patients responded (4 complete responses [CRs] and 57 partial responses [PRs]), and 37% of patients had stable disease. In the GC arm, 30% of patients responded (3 CRs and 54 PRs), and 33% of patients had stable disease (P = 0.213).

Table 2. Tumor Response Data

Tumor response	GC (n = 186 patients)		MIC/MVP (n = 186 patients)
	No.	%	No.	%
GC: gemcitabine and carboplatin; MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin; NOS: not otherwise specified.
Complete response	3	1	4	2
Partial response	54	29	57	31
Stable disease	61	33	69	37
Progressive disease	63	34	49	26
Died of toxicity	0	0	2	1
NOS	5	3	5	3

Patients with Stage IIIA NSCLC

We compared patients who had Stage IIIA disease with patients who had Stage IIIB and IV disease in terms of median survival and response rates (CRs and PRs). There were no significant differences at each stage between the two treatment arms for either variable (Table 3).

Table 3. Comparison of Patients with Stage IIIA and Stages IIIB and IV Nonsmall Cell Lung Carcinoma

Stage	Median survival (days)		Response rate (%)
	GC	MIC/MVP	GC	MIC/MVP
GC: gemcitabine and carboplatin; MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin.
IIIA	431	270	55	38
IIIB	259	304	25	41
IV	203	204	32	27

Toxicity

Toxicity data were available for 338 patients. In 34 patients who experienced rapid deterioration, some toxicity data are missing: Fifteen patients were lost to follow-up (2 patients had no treatment, and 13 patients did not attend for clinical assessment). In 19 instances, the patient clearly had deteriorated, and there was radiologic evidence to that effect, but the clinical annotation was inadequate for the extraction of toxicity data.

Episodes of nonhematologic, treatment-related toxicity are detailed in Table 4. Both regimens were tolerated well, with only mild-to-moderate toxicity. There was significantly more Grade 3 alopecia in the MIC/MVP arm, with a trend toward more Grade 1–2 flu-like symptoms and rash in the GC arm. There were no significant differences in the incidence of nausea and emesis, mucositis, peripheral neuropathy, renal toxicity, tinnitus, deafness, or edema.

Table 4. Nonhematologic Toxicity

Adverse event	Grade (%)a
	GC			MIC/MVP
	All (1–4)	3	4	All (1–4)	3	4
GC: gemcitabine and carboplatin; MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin. a Grading was performed according to the National Cancer Institute Common Toxicity Criteria. There was no significant difference between treatment arms with the exception of alopecia (P < 0.005).
Nausea/emesis	69	11	0	80	13	0
Alopecia	52	3	0	89	27	0
Mucositis	37	1	0	38	2	0
Rash	20	1	0	9	1	0
Peripheral neuropathy	20	1	0	17	1	0
Flu-like symptoms	33	1	0	11	0	0
Edema	21	2	0	17	1	0
Tinnitus
Intermittent	7	1	0	9	0	0
Continuous	3		0	1	0	0
Deafness
Intermittent	2	0	0	2	0	0
Continuous	4	1	0	1	0	0
Renal	5	0	1	11	1	0

The most frequent adverse events observed among patients in the GC arm were nausea/emesis (69%), alopecia (52%), mucositis (37%), flu-like symptoms (33%), and peripheral edema (21%). Among patients in the MIC/MVP arm, nausea/emesis (80%), alopecia (89%), and mucositis (38%) were most frequent adverse events.

The hematologic toxicities are summarized in Table 5. There was significantly more Grade 3–4 neutropenia (P = 0.001) and thrombocytopenia (P < 0.005) in the GC arm. Grade 3–4 anemia occurred in 15% of patients on both arms. Patients in the GC arm required significantly more blood transfusions (P < 0.005) and platelet transfusions (P < 0.005). There was a slightly significant difference between the i.v. antibiotic courses prescribed in the GC arm (P = 0.0489).

Table 5. Hematologic Toxicity

Toxicitya	No. of patients (%)		P valueb
	GC	MIC/MVP
GC: gemcitabine and carboplatin; MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin. a Grading was performed according to the National Cancer Institute Common Toxicity Criteria. b Chi-square test: P values for Grades 0–2 and Grades 3–4 are grouped together for GC arm and the MIC/MVP arm.
Neutropenia
Grade 3	55 (30)	38 (20)	—
Grade 4	37 (20)	26 (14)	0.001
Thrombocytopenia
Grade 3	46 (25)	10 (5)	—
Grade 4	49 (26)	12 (6)	0.001
Anemia
Grade 3	26 (14)	26 (14)	—
Grade 4	2 (1)	1 (1)	0.890
Platelet transfusions (no. of units)	316	49	—
Platelet transfusions (no. of patients)
No	116	162	—
Yes	54	12	0.005
Blood transfusions (no. of units)	371	235	—
Blood transfusions (no. of patients)
No	82	112	—
Yes	88	63	0.005
Intravenous antibiotics
No. of courses	40	25	—
No. of patients
No	140	157	—
Yes	30	17	0.049
Oral antibiotics
No. of courses	89	86	—
No. of patients
No	116	117	—
Yes	53	57	0.872

Although there were no dose reductions in the MIC/MVP arm, there were delays for investigation of deteriorating renal function and control of symptoms but not for low blood counts. For patients receiving gemcitabine, the Day 8 and Day 15 doses were adjusted according to the blood counts. There was a Day 8 gemcitabine dose reduction in 6% of patients over all cycles (0.5% in the first cycles increasing to 13% with Cycle 4). Omission of the Day 8 gemcitabine dose occurred in 8% of patients overall (range, 6–10% of patients). Dose reductions were more frequent for the Day 15 gemcitabine dose (13%, 12%, 13%, and 15% of patients for Cycles 1–4, respectively), for an overall rate of 13%. The gemcitabine dose was omitted on Day 15 for 51% of patients overall and for 51%, 51%, 59%, and 60% of patients for Cycles 1–4, respectively. Most of the omissions in the GC arm were due to hematologic toxicity, with other patients coming off protocol because of disease progression, deterioration in PS, and patient refusal. In total, patients received 615 cycles of MIC/MVP (average, 3.31 cycles per patient) and 587 cycles of GC (3.16 cycles per patient).

Patients with a PS of 2–3 did not have more hematologic or nonhematologic toxicity than patients with a PS of 0–1 (Table 6). Overall Grade 3–4 hematologic toxicity (neutropenia and thrombocytopenia) was less pronounced with MIC/MVP, but there was significantly less Grade 3 alopecia and a trend toward fewer of other platinum-associated, nonhematologic toxicities (emesis and renal toxicity) with GC.

Table 6. Hematologic Toxicity by WHO Performance Status

Toxicitya	No. of patients		P value
	PS 0–1	PS 2–3
WHO: World Health Organization; PS: performance status. a Grading was performed according to the National Cancer Institute Common Toxicity Criteria.
Anemia
Grade 1–2	203	80	0.191
Grade 3–4	34	21	—
Neutropenia
Grade 1–2	120	60	0.121
Grade 3–4	117	39	—
Thrombocytopenia
Grade 1–2	156	67	0.949
Grade 3–4	83	34	—

Disease-Related Symptoms

Symptoms were assessed by physicians, as in other studies, at baseline and at 4 weeks postchemotherapy. Data were available for 50% of patients, with the remaining data missing because patients decided to terminate treatment, because patients were too unwell for evaluation, or because the results were not documented. The most pronounced subjective improvement was in cough, with 43% of patients in the GC arm and 36% of patients in the MIC/MVP arm achieving a better score (Table 7). There was also an improvement in hemoptysis in 12% of patients in the GC arm and 19% of patients in the MIC/MVP arm. Chest pain diminished in 27% of patients on each arm. Dyspnea responded in 31% of patients in the GC arm and 26% of patients in the MIC/MVP arm and worsened in 34% of patients in the GC arm and 33% of patients in the MIC/MVP arm. Fatigue was worse in 43% of patients in the GC arm and 49% of patients in the MIC/MVP arm (Table 7).

Table 7. Disease-Related Symptoms (Percentage of patients)

Symptom	Percent change
	GC (%)			MIC/MVP (%)
	Better	Same	Worse	Better	Same	Worse
GC: gemcitabine and carboplatin; MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin; SVCO: superior vena cava obstruciton.
Cough	43	39	18	36	41	23
Hoarseness	6	86	8	9	80	11
Hemoptysis	12	86	2	19	79	2
Dyspnea	31	35	34	26	41	33
SVCO	1	98	1	0	99	1
Stridor	0	99	1	1	99	0
Chest pain	27	56	17	27	60	13
Pain elsewhere	23	61	16	8	72	20
Constipation	21	68	11	13	76	11
Anorexia	26	49	25	17	60	23
Dysphagia	7	87	6	2	89	9
Depression	5	84	11	7	77	16
Fatigue	17	40	43	15	36	49

PS

Data were available on PS at baseline and at 4 weeks for 77% of patients and for baseline, 4 weeks, and 8 weeks after therapy for 41% of patients. Some patients had died within 4 weeks of their previous chemotherapy or were not assessable at these times. For assessable patients, changes were comparable in both arms, with 16% showing an improved PS and 43% remaining unchanged.

Hospitalization

We considered the number of emergency overnight stays for the 343 patients who were treated at the Manchester hospitals. There were 617 overnight stays among patients in the GC arm and 658 overnight stays among patients in the MIC arm (P < 0.005) (Table 8).

Table 8. Emergency Admissions

Variable	GC	MIC
GC: gemcitabine and carboplatin; MIC: mitomycin, ifosfamide, and cisplatin. a P < 0.005.
No. of patients	169	174
No. of possible nights	14,958	11,840
No. of actual cycles given by interval	554 × 27	592 × 20
No. of overnight admissions	617	658a
Percentage of nights	4.12	5.56

QoL

Sixty percent of patients in the GC arm and 50% of patients in the MIC/MVP arm completed QoL questionnaires both at baseline and after Cycle 2, whereas 29% and 27% of patients, respectively, completed end-of-treatment forms. Because small differences were unlikely to reflect clinical benefit, interest was focused on those subscales or symptom items that showed a > 10% between-treatment difference in the number of patients who improved or deteriorated. Findings for ease of presentation are displayed as histograms in Figures 3 and 4.

Figure 3. European Organization for Research and Treatment of Cancer (EORTC) QLC-C30 and LC13 subscales and items that showed ≥ 10% between-treatment differences in the proportion of patients who reported improvement and deterioration from baseline to midcycle. MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin; GC: gemcitabine and carboplatin.

Figure 4. European Organization for Research and Treatment of Cancer (EORTC) QLC-C30 and LC13 subscales and items that showed ≥ 10% between-treatment differences in the proportion of patients who reported improvement and deterioration from baseline to the end of the study. MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin; GC: gemcitabine and carboplatin.

At midtreatment (after Cycle 2), of the 25 variables analyzed, 6 variables demonstrated differences > 10% between the 2 arms. All improvements were greater for the MIC/MVP arm compared with the GC arm. The improved variables were concentration, nausea/emesis, constipation, and cough. Two variables showed deterioration and were worse in the MIC/MVP arm compared with the GC arm: emotional well-being and alopecia.

At the end of treatment, five variables had improved: Four were better in the GC arm (average score, work/hobbies, chest pain, and other pain), whereas dyspnea had improved more in the MIC/MVP arm. Eleven variables had deteriorated, and 9 of these were worse for the MIC/MVP arm (global health, functional, social, fatigue, nausea/emesis, constipation, dyspnea, alopecia, and chest pain), and 2 were worse for the GC arm (sleeping and appetite). Overall, although the midtreatment scores seemed to favor MIC/MVP, the end-of-treatment QoL indicated that GC was the better regimen. The HADS data did not show any significant differences between baseline scores and midcycle or end-of-treatment scores for either arm (Table 9).

Table 9. Hospital Anxiety and Depression Scale Scores

HADS score	No. of patients (%)
	Baseline		Midtreatment		Baseline		Posttreatment
	MIC/MVP	GC	MIC/MVP	GC	MIC/MVP	GC	MIC/MVP	GC
HADS: Hospital Anxiety and Depression Scores; MIC: mitomycin, ifosfamide, and cisplatin; MVP: mitomycin, vinblastine, and cisplatin; GC: gemcitabine and carboplatin.
Anxiety subscale
No. of patients	81	97	81	97	57	62	58	62
< 11	55 (68)	79 (81)	63 (78)	88 (91)	44 (77)	54 (87)	45 (78)	54 (87)
≥ 11	26 (32)	18 (19)	18 (22)	9 (9)	13 (23)	8 (13)	13 (12)	8 (13)
Depression subscale
No. of patients	80	96	79	97	57	58	46	54
< 11	72 (90)	88 (92)	69 (87)	83 (86)	55 (96)	58 (95)	46 (81)	54 (87)
≥ 11	8 (10)	8 (8)	10 (13)	14 (14)	2 (4)	3 (5)	11 (19)	8 (13)

There was no indication that patients who had a PS of 2–3 were less likely to complete their QoL and HADS questionnaires compared with patients who had a PS of 0–1: Eighty percent of patients with a PS of 0–1 and 89% of patients with a PS of 2–3 completed baseline QoL questionnaires, whereas 62% of patients in both PS categories completed questionnaires after Cycle 1 (Fig. 5).

Figure 5. Quality of life and Hospital Anxiety and Depression Score completion rates by performance status (PS).

Subsequent Treatment

Radiotherapy was given to 101 patients (59%) from the GC arm and 99 patients (53%) from the MIC/MVP arm. Second-line chemotherapy was administered to 13 patients (7%) who had previously received GC and 8 patients (4%) who had previously received MIC/MVP. Neither of these interventions had statistical significance because the numbers were too small. No patients in the trial underwent surgery after chemotherapy.

DISCUSSION

Randomized trials and several meta-analyses have demonstrated that chemotherapy for patients with advanced NSCLC who have a good PS increases survival, relieves symptoms, and may improve QoL.3, 4 The advent of newer agents, including gemcitabine, has led to investigations to determine the most useful regimen. Earlier randomized trials of these newer agents versus best supportive care alone and later combination chemotherapy trials led to their approval by the National Institute of Clinical Excellence based on efficacy and cost effectiveness.5 In the current trial, carboplatin was combined with gemcitabine in an attempt to minimize toxicity. It has been shown that the older regimens (MIC and MVP) are superior to combined cisplatin and etoposide in terms of response rate (40%, 36%, and 23%, respectively) and median survival (36 weeks, 42 weeks, and 27 weeks, respectively).31 The current trial showed no statistically significant differences between the two treatment arms in terms of survival, time to disease progression, or response rates.

Recently, there has been increasing interest in reporting additional endpoints of patient benefit, symptom relief, and change in QoL. Indeed, improvements in patient's symptoms can be considerable and may exceed improvements in the objective tumor response rate. For MVP, MIC, and single-agent gemcitabine, hemoptysis can be relieved in 60% of patients, cough can be relieved in > 40% of patients, pain can be relieved in 30% of patients, and dyspnea can be relieved in 25% of patients.2, 4, 27 There also can be an improvement in PS and reduced requirements for palliative radiotherapy.4, 27 Gemcitabine, an agent with a favorable toxicity spectrum, has been combined with cisplatin (GP), and the GP combination was superior to cisplatin alone (odds ratio [OR], 30% vs. 11%, respectively), with a median survival of 9.1 months versus 7.6 months, respectively.32 However, a comparison of GP with MIC showed an increased OR of 38% versus 26%, respectively, with no significant difference in survival (median, 8.6 months vs. 9.6 months, respectively) or QoL.12 Those two trials also went further and investigated the impact of hematologic toxicity, as graded conventionally from laboratory tests, on the patient's transfusion requirements and the incidence of febrile neutropenia.

These data are displayed together with the current findings (Table 10). However, the differences may have been due to more frequent blood counts that were taken in the GP arm compared with the MIC/MVP arm or with patients who received single-agent cisplatin. The transfusion guideline triggers in the current study were platelets < 20 × 10⁹/L and hemoglobin < 9.5 g/dL but were not stated in the other two trials. All three trials reported a reduction or an omission of gemcitabine, particularly on Day 15, in approximately 50% of patients, mainly due to thrombocytopenia.12, 32 More recently, the scheduling of gemcitabine has changed to Day 1 and Day 8 in combination with platinum with similar dose intensity, response rates, and survival.17, 33, 34 Preliminary data from a Phase III trial35 of GC on a 3-weekly schedule compared with MIC, with gemcitabine 1.2 g on Days 1 and 8 and carboplatin AUC 5, has shown a significant increase in survival of 10.2 months in the GC arm and 6.9 months in the MIC arm, compared with 8.3 months and 7.9 months, respectively, in our trial. The 3-weekly schedule appears to produce less hematologic toxicity, with Grade 3–4 thrombocytopenia in 25% of patients in the GC arm and 7% of patients in the MIC arm, compared with 51% of patients in our GC 4-weekly arm. This further strengthens the case for the 3-weekly scheduling.

Table 10. Hematologic Toxicity and Transfusion Requirements for Red Cells, Platelets, and Febrile Neutropenia

Toxicity	Percentage with hematalogic toxicity (transfusion requirements)
	Crino et al.12		Sandler et al.32	Current study
	GP	MIC	GP	GC	MIC/MVP
GP: gemcitabine and cisplatin; MIC: mitomycin, ifosfamide, and cisplatin; GC: gemcitabine and carboplatin; MVP: mitomycin, vinblastine, and cisplatin.
Anemia	31 (23)	25 (19)	25 (38)	15 (52)	15 (36)
Neutropenia	40 (1)	34 (0)	57 (5)	50 (18)	34 (10)
Thrombocytopenia	64 (15)	28 (3)	50 (20)	51 (34)	11 (7)

The cisplatin dose in our MIC/MVP arm was somewhat lower compared with the dose used in many cisplatin-containing chemotherapies. Nonetheless, our results bear comparison with those achieved elsewhere, taking into account the high proportion of patients with a PS of 2 in our study. For example, in a randomized, Phase II study9 that compared GP with cisplatin dosed at 70 mg/m² and 100 mg/m², the higher dose produced no statistically significant benefit in median survival, which was 12 months for the whole trial. Although it was better compared with the median survival of 8 months seen in our trial, only 10% of the patients enrolled in that study had a PS of 2 compared with > 30% of patients in our trial.

PS is an independent predictor of survival in patients with advanced NSCLC.36, 37 Previously it was found that rapid deterioration and poor compliance prevented QoL data collection and subsequent analysis in patients with a poor PS.8, 38 Thus, many trials exclude patients with a PS of 2 or 3. However, although survival and response rates may not be altered, we also should consider any symptomatic benefit, because this is an important consideration.4 It is noteworthy that, in the ECOG study17 that compared the standard of cisplatin and paclitaxel with either gemcitabine plus cisplatin, cisplatin plus docetaxel, or carboplatin plus paclitaxel, there was no difference in survival, with a median survival of 7.9 months and a 2-year survival rate of 11%, similar to the current trial with GC. However, in the ECOG study, after 66 patients with a PS of 2 had been enrolled, recruitment was stopped because of a higher rate of toxicity. We did not observe this level of toxicity in the larger patient cohort in the current study, in which the median survival for patients with a PS of 2 was 6.2 months, compared with 3.9 months in the ECOG trial.17, 39 In the current study, these patients were as likely to complete the QoL questionnaires as the patients who had a better PS. The patients with a poor PS in the current study may have tolerated the chemotherapy so well because of the lower cisplatin dose used. There is a clear need to investigate further treatment options for patients with an impaired PS who currently tend to be excluded from clinical trials. This view was supported recently by a large trial40 of single-agent paclitaxel versus combination paclitaxel plus carboplatin in patients with advanced NSCLC that confirmed the finding that patients who had a PS of 2 fared worse compared with patients who had a PS of 0–1 but also showed that patients who had a PS of 2 survived longer on the combination treatment than on single-agent therapy (4.7 months vs. 2.4 months, respectively; P = 0.0177).

In general, the QoL data collected in the current study favored the GC combination. This may have been influenced by our policy of admitting patients on MIC/MVP for each cycle of treatment. By completion of treatment, there was a greater deterioration among patients in the MIC/MVP arm in terms of global health, function, social activities, fatigue, nausea/emesis, dyspnea, alopecia, chest pain, and constipation reported by the EORTC subscale (9 of 11 variables). In a comparison of the two-drug regimen GP with single-agent cisplatin, there was no overall difference in QoL, although deterioration in physical and functional well-being and total FACT-L scores was described for the doublet.32 In the Italian study, both treatments were associated with a moderate decrease in physical functioning and worsening fatigue. However, global QoL was no different for MIC or GP, and improvements were noted in pain and cough.12 The findings of the current study support other data describing symptom benefits with gemcitabine in a study that included patients who had a PS > 2.27 In addition, any intervention that reduces the period of hospitalization, such as the GC regimen in the current trial, is particularly pertinent when survival tends to be short.

In conclusion, this study did not reveal any statistically significant differences between GC and MIC/MVP. Hematologic toxicity and the need for transfusion and antibiotic support were lower for patients in the MIC/MVP arm, but GC involved less hospitalization for emergency treatment. Overall, gemcitabine combined with carboplatin is an acceptable alternative to cisplatin-based MIC/MVP.