The impact of age on toxicity, response rate, quality of life, and survival in patients with advanced, Stage IIIB or IV nonsmall cell lung carcinoma treated with carboplatin and paclitaxel

Abstract

BACKGROUND

The optimal treatment strategy for elderly patients with advanced nonsmall cell lung carcinoma has not been defined to date. The authors performed a retrospective analysis of a Phase III trial that treated patients who had Stage IIIB or IV nonsmall cell lung carcinoma with carboplatin and paclitaxel and analyzed the impact of age on response rate, survival, toxicity, and quality of life.

METHODS

Patients with Stage IIIB or IV NSCLC were randomized to receive either 4 cycles of carboplatin at an area under the curve (AUC) of 6 and paclitaxel at a dose of 200 mg/m² every 21 days or treatment with carboplatin and paclitaxel (C/P) until they developed disease progression. At the time of disease progression, all patients on both arms were to receive second-line weekly paclitaxel at a dose of 80 mg/m² per week. In this analysis, patients age 70 years and older were compared with patients younger than age 70 years. In addition, a minimum log rank P value analysis was performed in an attempt to identify other potential age splits that may have been significant.

RESULTS

Two hundred thirty patients were randomized. Sixty-seven patients were age 70 years or older (29%). The median number of cycles delivered for both age groups was 4 cycles (range, 0–19 cycles). No statistically significant differences in any of the most common toxicities (Grade ≥ 2) associated with C/P were identified (data from Cycles 1–4) for patients younger than age 70 years compared with patients age 70 years and older, respectively, including neutropenia (38% vs. 35%), neuropathy (13% vs. 16%), leukopenia (7% vs. 13%), myalgia/arthralgia (15% vs. 9%), malaise (8% vs. 15%), anemia (9% vs. 4%), thrombocytopenia (7% vs. 9%), anorexia (8% vs. 4%), and nausea/emesis (14% vs. 15%). In addition, no potential age splits that may have been significant were found using a minimum log rank P value analysis.

CONCLUSIONS

The current analysis demonstrated that C/P exhibited similar toxicity profiles in patients age 70 years and older compared with patients younger than age 70 years. The survival rates were not different between the two age groups, and there was no difference in progression of quality-of-life outcomes. In fit, elderly patients, C/P represented a reasonable standard regimen. Cancer 2003;98:779–88. © 2003 American Cancer Society.

Nonsmall cell lung carcinoma (NSCLC) remains the leading cause of cancer mortality in the United States. The optimal treatment for elderly patients with advanced NSCLC remains undefined. This is becoming an increasingly relevant issue as the demographics of this disease change. According to a report from the National Cancer Data Base, from 1985 to 1995, the proportion of patients age 70 years and older with lung carcinoma increased from 35% to 43%.1 Approximately 80% of these patients will have carcinoma with nonsmall cell histology, and > 30% will present with Stage IV disease.1 Although chemotherapy is offered increasingly to patients with advanced-stage disease (Stage IIIB or IV), older patients still are less likely to be offered this modality.1–4 Both patient bias and physician bias likely contribute to this discrepancy.5 Elderly patients frequently have been excluded from prospective trials, complicating the ability to generalize positive results to this population.6 In addition, several physiologic changes associated with aging may diminish the ability of elderly patients to tolerate chemotherapy and thereby reduce the therapeutic index of these agents.7, 8 However, it has not been shown that advanced age is a consistent prognostic factor for patients with Stage IV disease.9 Furthermore, there is now prospective, randomized data showing that selected elderly patients who are treated with chemotherapy may achieve comparable survival and quality-of-life (QoL) benefits compared with younger patients.10–12

Platinum-based combination regimens remain standard front-line therapy for most patients with advanced-stage NSCLC. Based on the meta-analysis reported by the Nonsmall Cell Lung Cancer Collaborative Group, cisplatin-based therapy was associated with an absolute improvement in the 1-year survival rate of 10% (range, 5–15%) over best supportive care.13 Since that analysis, most modern regimens have combined a third-generation chemotherapy agent, including paclitaxel, docetaxel, vinorelbine, or gemcitabine, with either cisplatin or carboplatin. Based on several randomized trials, the efficacy of these regimens appears to be equivalent, with a median survival of 8–10 months and 1-year and 2-year survival rates of 30–40% and 10–15%, respectively.14–16

In a large, randomized trial conducted by the Eastern Cooperative Oncology Group (ECOG), there were no differences in response rates or overall survival among four common, front-line, platinum-based regimens, including cisplatin and paclitaxel, cisplatin and gemcitabine, cisplatin and docetaxel, and carboplatin and paclitaxel. The latter combination was associated with decreased toxicity compared with the other regimens.14 The carboplatin/paclitaxel combination has become the most common regimen used in the United States for the front-line treatment of patients with advanced NSCLC.17 The toxicities associated with the carboplatin and paclitaxel regimen have been described14, 16, 18 well; however, there are scant published data on the effect of this combination in the elderly population. Therefore, in an effort to assess the impact of age on the efficacy and toxicity of the carboplatin and paclitaxel regimen, we conducted an analysis of data from a Phase III trial that randomized 230 patients with advanced (Stage IIIB/IV) NSCLC to a defined duration of therapy (4 cycles) with carboplatin and paclitaxel versus continuous therapy until they developed progressive disease. The results of this study were reported previously by Socinski and colleagues and showed no survival or QoL benefit when treatment was extended beyond four cycles.19

MATERIALS AND METHODS

Patient Selection

The patients who were included in the current analysis were initially part of a multicenter, Phase III trial that addressed the optimal duration of therapy in patients with advanced NSCLC. For that trial, patients were required to have histologic or cytologic proof of Stage IIIB or Stage IV NSCLC. Among patients with Stage IIIB disease, those selected were not candidates for combined-modality therapy (due to malignant pleural effusions or advanced supraclavicular adenopathy). Patients could have measurable or evaluable disease and could not have received prior chemotherapy for NSCLC. Prior radiotherapy was allowed; however, patients had to be at least 1 week from the completion of radiotherapy. Patients with treated brain metastases were allowed. Patients were required to have a Karnofsky performance status (PS) ≥ 70% and adequate end-organ function, which was defined as an absolute neutrophil count ≥ 1500/mm³, platelets ≤ 100,000/mm³, creatinine < 2.0 × the institutional upper normal limit (IUNL), bilirubin < 1.5 × IUNL, aspartate aminotransferase < 2.5 × IUNL, and no clinically significant baseline peripheral neuropathy. All patients were required to participate in the QoL component of the study and were required to provide written, informed consent. This trial was reviewed by the Protocol Review Committee of the Lineberger Comprehensive Cancer Center (LCCC) and by the Institutional Review Board of the University of North Carolina School of Medicine (as well as all participating centers) and was labeled LCCC 9719.

Treatment Plan and Response Assessment

In the Phase III trial, patients were randomized to either Arm A or Arm B. Patients randomized to Arm A received four cycles of carboplatin and paclitaxel, whereas patients on Arm B received carboplatin and paclitaxel until disease progression. On both arms, carboplatin was given at an area under the curve (AUC) of 6 (using the Calvert Equation), and paclitaxel was given at a dose of 200 mg/m² over 3 hours every 21 days. Standard premedications for paclitaxel were administered and included dexamethasone 20 mg intravenously (IV), diphenhydramine 50 mg, and either ranitidine 50 mg IV or cimetidine 300 mg IV given 30 minutes prior to paclitaxel.

Because the distribution of the number of cycles of carboplatin and paclitaxel was not significantly different between Arm A and Arm B (P = 0.22; Wilcoxon rank sum test using normal scores) of the Phase III trial, the database was analyzed based on age (younger than age 70 years or age 70 years and older) independent of the arm to which the patients were randomized. Patients were assessed for disease response every 2 cycles or every 6 weeks of first-line therapy. Standard criteria for response to chemotherapy were used. Patient response to treatment was categorized as either complete, partial, stable or progressive according to these criteria. Reasons for treatment discontinuation on the Phase III trial were completion of four cycles, excessive toxicity, disease progression, and when deemed in the best interest of the patient. The Phase III trial had predetermined second-line (SL) therapy, which was low-dose weekly paclitaxel (80 mg/m²). Patients were to receive this therapy at the time of progression assuming they met eligibility criteria similar to those mentioned above with one exception: Karnofsky PS ≥ 60% rather than 70%. The details of SL weekly paclitaxel are reported elsewhere.20 Patients were treated with weekly paclitaxel until the developed disease progression. The first assessment for disease progression occurred after 8 weeks of therapy. After this point, 1–2-week treatment breaks were allowed at the investigator's discretion.

QoL

The Functional Assessment of Cancer Therapy-Lung (FACT-L) questionnaire was used to evaluate the QoL endpoint in this study: The FACT subscale that measures QoL concerns related to neurotoxicity and taxane toxicity (NTTX) also was administered. The FACT-L was chosen because of its comprehensive coverage of QoL issues in patients with lung carcinoma, its strong psychometric properties, and its widespread use in cooperative group trials. The Trial Outcome Index–Lung (TOI-L), which combines the scores from the physical, functional, and lung carcinoma specific well-being subscale scores of the FACT-L, was chosen as the primary QoL endpoint: The secondary QoL endpoint was the TOI-NTTX, which combines the physical, functional, and NTTX subscale scores. All patients were required to complete the FACT-L at the time of registration onto the trial. Once this was completed, subsequent QoL assessments were conducted by telephone at 5 weeks, 11 weeks, and 25 weeks after the start of chemotherapy. This timing was chosen for several reasons: 1) because similar assessment times were used in Protocol E5592 of the Eastern Cooperative Oncology Group, the results of which were released immediately before the initiation of this trial; 2) to obtain QoL information close in time to the evaluation of response to treatment after Cycles 2 and 4 (Weeks 6 and 12) but prior to the time patients received the results of their evaluations to avoid potential QoL effects of that knowledge; 3) to allow the evaluation of QoL at a time long enough after the end of the defined duration of chemotherapy so that treatment side effects would have diminished in patients who received only 4 cycles.

Statistical Design

In the decision analysis regarding sample size for the Phase III trial,19 an exponential survival distribution was assumed. A difference in 1-year survival between the two arms of 15% was used. Given an α of 0.05, a power of 0.80, an accrual period of 18 months, and a follow-up of 24 months after the last patient was accrued, the total sample size required was 206 patients (103 patients per arm). A 10% nonevaluable rate was assumed, making the accrual goal 230 patients. The analysis of efficacy was performed on all patients who were randomized according to the intent-to-treat principle. These 230 patients were then analyzed based on age and exploring clinical outcomes of interest in this population, as noted below.

Age 70 years has become a common age split to define the elderly population. Therefore, in the current analysis, patients age 70 years and older were compared with patients younger than age 70 years. Baseline patient demographics, including stage, PS, gender, histology, and treatment group, as well as number of treatment cycles, number of patients who received SL therapy, response rate (overall response, comprised of complete responses and partial responses), reasons for treatment discontinuation, and hematologic and nonhematologic toxicity were compared and tested using the Fisher exact test. Because a median of four treatment cycles was the same for both age groups, the incidence of both hematologic and nonhematologic toxicity was compared for Cycles 1–4 only. The survival curves were estimated using the Kaplan–Meier method and were compared using the log rank test. In an effort to identify additional potential age splits that may be significant, a minimum log rank P value analysis was conducted. The QoL data were analyzed with a linear mixed model, which allowed the inclusion of all available longitudinal data.

RESULTS

Patient Demographics

Between January 1998 and January 2000, 230 patients were entered onto the Phase III trial and served as the database for the current analysis. Patients were enrolled from both academic (60%) and community cancer centers (40%). The majority of patients were from the southeastern United States (68%), with the remainder from the Midwest (28%) and northeast (4%) regions of the country. Table 1 shows the demographics for all patients entered onto the trial as well as the distribution based on age (younger than age 70 years and age 70 years and older). The median age for all patients was 63 years (range, 31–82 years). The majority of the patients were male (63%), had Stage IV disease (87%), and had disease with nonsquamous histology (78%). There were 67 patients (29%) who were age 70 years or older, with 24 patients (10%) age 75 years or older and 7 patients (3%) age 80 years or older. We found no differences in disease stage, gender, or histology between the two groups, although there were more patients with poor PS (70–80%) in the elderly group (63% vs. 42%; P = 0.005). In addition, more elderly patients were randomized to Arm B (treatment until disease progression) rather than arm A (4 cycles) of the Phase III trial (64% vs. 36%; P = 0.007).

Table 1. Patient Demographics

Characteristic	No. of patients (%)
	All patients (n = 230)	Younger than age 70 yrs (n = 163)	Age 70 yrs and older (n = 67)
a P = 0.005. b P = 0.007.
Gender
Male	144 (63)	98 (60)	46 (69)
Female	86 (37)	65 (40)	21 (31)
Stage
IIIB	30 (13)	20 (12)	10 (15)
IV	200 (87)	143 (88)	57 (85)
Histology
Squamous	50 (22)	30 (18)	20 (30)
Nonsquamous	180 (78)	133 (82)	47 (70)
Performance statusa
90–100	119 (52)	94 (58)	25 (37)
70–80	111 (48)	69 (42)	42 (63)
Treatment groupb
Arm A (four cycles)	114 (50)	90 (55)	24 (36)
Arm B (until progression)	116 (50)	73 (45)	43 (64)

Treatment Administration Summary

Table 2 displays the treatment administration. The median number of cycles for both age groups was 4 (range, 0–19 cycles). However, we identified significant differences in the reasons for treatment discontinuation between the two age groups (Table 3). More elderly patients discontinued treatment due to patient or physician choice (37% vs. 19% for patients age 70 years and older vs. patients younger than age 70 years, respectively; P = 0.006), whereas more patients in the younger group discontinued treatment after they completed therapy as planned (16% vs. 32%, respectively; P = 0.02). This difference was due in part to treatment arm assignment (Table 3); when patients in Arm A (≤ 4 cycles only) were considered separately, the reasons for treatment discontinuation did not differ significantly by age group. In contrast, when patients in Arm B (treatment until disease progression) were considered, the reasons for treatment discontinuation were significantly different between the two age groups (P = 0.017; overall Fisher exact test). Within Arm B, patient or physician choice remained the predominant reason that elderly patients stopped therapy (51%), followed by toxicity (21%), disease progression (21%), and death (7%). In the younger group, disease progression was the predominant reason patients stopped therapy (47%), followed by patient or physician choice (27%), toxicity (14%), and death (11%).

Table 2. Treatment Administration

Variable	Younger than age 70 yrs (n = 163)	Age 70 yrs and older (n = 67)
a P = 0.006.
No. of cycles
Median	4	4
Range	0–19	1–11
No. of patients who received second-line therapy (%)	81 (50)	20 (30a)
Number
Percent

Table 3. Reasons for Treatment Discontinuation by Treatment Arm and Age Group

Reason off study	Arm Aa (%)		Arm B (%)
	Younger than age 70 yrs (n = 90)	Age 70 yrs and and older (n = 24)	Younger than age 70 yrs (n = 73)b	Age 70 yrs and older (n = 43)
a Overall Fisher exact test comparing the reasons for going off study by age group within each treatment arm (Arm A, P = 0.55; Arm B, P = 0.017). b Includes one patient who was coded as still on treatment.
Completed therapy (Arm A patients)	58	46	0	0
Disease progression	17	17	47	21
Toxicity	3	4	14	21
Patient/physician choice	12	12	27	51
Death	10	21	11	7

Treatment with SL therapy is outlined in Table 2. Overall, 101 patients (44%) in the Phase III trial were treated with SL therapy, including 62 patients who received weekly paclitaxel. The remaining patients received alternative regimens or single agents, the most common of which was gemcitabine. A greater proportion of patients younger than age 70 years received SL therapy (50% vs. 30%; P = 0.006).

Toxicity

The overall rates of Grade 1–4 hematologic toxicity were comparable between the two groups (Table 4). There was an increase in Grade 3–4 leukopenia in the elderly group (13% vs. 7%), but this was not statistically significant (P = 0.12). Four patients experienced febrile neutropenia—three in the younger group and one in the older group. Table 5 displays the episodes of Grade 1–4 nonhematologic toxicity. A greater proportion of elderly patients experienced malaise and fatigue (39% vs. 24%; P = 0.03). However, when Grade 3 and Grade 4 toxicity was considered separately, the difference was no longer statistically significant (5% vs. 1% for patients age 70 years and older vs. patients younger than age 70 years, respectively; P = 0.15). It also should be noted that the typical toxicities associated with this regimen (i.e., neuropathy and myalgia/arthralgia) did not occur more commonly in elderly patients compared with younger patients.

Table 4. Distribution of the Most Common Hematologic Toxicities Associated with Carboplatin and Paclitaxel by Grade and Age

Toxicity	No. of patients(%)							P value
	All grades (1–4)			Grade 3		Grade 4
	Younger than age 70 yrs	Age 70 yrs and older	P value	Younger than age 70 yrs	Age 70 yrs and older	Younger than age 70 yrs	Age 70 yrs and older
Neutropenia	76 (47)	40 (60)	0.08	39 (24)	13 (19)	23 (14)	11 (16)	0.77
Leukopenia	45 (28)	27 (40)	0.06	10 (6)	7 (10)	1 (1)	2 (3)	0.12
Anemia	105 (64)	39 (58)	0.45	12 (7)	3 (4)	2 (1)	0 (0)	0.41
Thrombocytopenia	55 (34)	23 (34)	1.0	7 (4)	4 (6)	4 (2)	2 (3)	0.58

Table 5. Distribution of the Most Common Nonhematologic Toxicities Associated with Carboplatin and Paclitaxel by Grade and Age

Toxicity	No. of patients(%)							P value
	All grades (1–4)			Grade 3		Grade 4
	Younger than age 70 yrs	Age 70 yrs and older	P value	Younger than age 70 yrs	Age 70 yrs and older	Younger than age 70 yrs	Age 70 yrs and older
Neuropathy	64 (39)	29 (43)	0.66	7 (5)	6 (9)	0 (0)	0 (0)	0.54
Nausea/emesis	53 (33)	20 (30)	0.76	7 (4)	4 (6)	2 (1)	0 (0)	0.84
Myalgia/arthralgia	59 (36)	18 (27)	0.22	3 (2)	2 (3)	0 (0)	0 (0)	0.29
Anorexia	24 (15)	8 (12)	0.68	2 (1)	0 (0)	0 (0)	0 (0)	0.41
Malaise/fatigue	39 (24)	26 (39)	0.03	1 (1)	3 (4)	0 (0)	1 (1)	0.15

QoL

QoL data were available for 218 patients at baseline: (n = 164 [95.6%] younger than age 70 years; n = 66 [99%] age 70 years and older). Data completion rates and reasons for missing data did not differ between the two age groups (P values: range, 0.08–0.90 for baseline to 25 week assessments). At baseline, QoL did not differ between the two groups (TOI-L, P = 0.70; TOI-NTTX, P = 0.89). To determine whether the change in QoL over time differed by age group, a mixed model was run with two factors—age category and assessment point—and an interaction term (age category × assessment point). An unstructured covariance pattern for correlations within patients was used because it allows the variance of the endpoint to be different at each visit and the covariance of each pair of visits to be different. There was a significant effect of assessment point (P < 0.0001 for both TOI-L and TOI-NTTX), indicating that QoL changed over time. However, neither the main effect of age (TOI-L, P = 0.73; TOI-NTTX, P = 0.90) nor the interaction term (TOI-L, P = 0.49; TOI-NTTX, P = 0.42) was significant. Thus, QoL did not differ between patients age 70 years and older and patients younger than age 70 years, nor did the two groups demonstrate a differential rate of change over time. Table 6 shows the means scores adjusted for the factors in the model.

Table 6. Least-Squares Mean Estimates ± Standard Errors for Quality of Life by Age Group and Assessment Time

Outcome measure	Baseline		5 wks		11 wks		25 wks
	Younger than age 70 yrs	Age 70 yrs and older	Younger than age 70 yrs	Age 70 yrs and older	Younger than age 70 yrs	Age 70 yrs and older	Younger than age 70 yrs	Age 70 yrs and older
TOI-L: Trial Outcome Index–Lung Cancer; TOI-Tax: Trial Outcome Index–Taxane. a Age effect: F(1,219) = 0.01 (P = 0.90); time effect: F(3,219) = 39.47 (P < 0.0001); age × time effect: F(3,219) = 0.94 (P = 0.42). b Age effect: F(1,219) = 0.12 (P = 0.73); time effect: F(3,219) = 11.94 (P < 0.0001); age × time effect: F(3,219) = 0.81 (P = 0.49).
TOI-La	56.6 ± 1.1	57.2 ± 1.6	53.3 ± 1.2	55.5 ± 1.8	51.3 ± 1.2	50.9 ± 1.9	50.5 ± 1.9	50.8 ± 2.7
TOI-Taxb	94.2 ± 1.3	93.9 ± 2.0	86.1 ± 1.4	88.4 ± 2.2	79.1 ± 1.6	78.9 ± 2.5	80.6 ± 2.3	77.7 ± 3.3

Response and Survival

Table 7 shows the response data. There were 2 patients age 70 years and older who achieved a complete response (3%) and 16 patients who achieved a partial response (24%), for an overall response rate of 27%. In the younger group, 3 patients achieved a complete response (2%), and 30 patients achieved a partial response (18%), for an overall response rate of 20%. This difference was not significant (P = 0.28).

Table 7. Distribution of Best Overall Response Rate by Age

Response	Younger than age 70 yrs (%)	Age 70 yrs and older (%)
a P = 0.28.
Overall responsea	20	27
Complete	2	3
Partial	18	24
Stable disease	37	33
Disease progression	28	18
Unknown	14	21

Eleven of 230 patients remained alive as of April 2003. The median follow-up for these patients was 45 months (range, 35–61 months). The median time to disease progression was longer for the elderly patients (4.8 months vs. 3 months; P = 0.049; Wilcoxon rank sum test), although this did not translate into an improvement in overall survival. Survival was estimated for both age groups from the start of first-line treatment. The median survival in the younger group was 7.8 months (95% confidence interval [95% CI], 6.3–9.1 months), compared with 7.1 months for the elderly patients (95% CI, 4.8–11.6 months). The actual 1-year and 2-year survival rates for patients younger than age 70 years and patients age 70 years and older were 30% (95% CI, 23–37%) and 14% (95% CI, 9–19%) versus 33% (95% CI, 22–44%) and 7% (95% CI, 1–14%), respectively (P = 0.65). Figure 1 shows the overall survival curves for the two age groups.

Figure 1. Survival by age group (patients younger than age 70 years and patients age 70 years and older). The inset shows the median survival in months and the 1-year and 2-year survival rates with 95% confidence intervals (95% CI).

Potential Age Splits

Figure 2 shows the data from the minimum log rank P value analysis. Different age splits were analyzed from age 45 years to age 75 years, in 5-year increments. There was no other age split identified that had a significant impact on survival at P < 0.05, even though a lower P value was required in this analysis due to the multiple comparisons performed.21

Figure 2. P values for survival based on age in years at diagnosis. The numbers of patients who were younger than the age identified are indicated.

SL Therapy

Of 62 patients who received SL weekly paclitaxel, 48 patients were younger than age 70 years, and 14 patients were age 70 years or older. The overall response rate was 8% (all partial responses), and there was no difference based on age (8.3% vs. 7.1% of patients younger than age 70 years and patients age 70 years and older, respectively; P = 0.53). In addition, there were no significant differences in the time to progression (2.1 months vs. 2.4 months, respectively; P = 0.98) or in the reasons for treatment discontinuation (P = 0.39) between the two age groups. Overall, disease progression (64% of patients) was the most common reason that SL therapy was stopped, followed by patient or physician choice (18% of patients), and death (13% of patients).

DISCUSSION

The management of advanced NSCLC in the elderly remains a difficult and complex issue. There are several physiologic and functional changes that accompany aging and can complicate the ability of elderly patients to tolerate chemotherapy.7, 8 In addition, the risk of certain comorbidities also has an impact on the ability of the elderly to tolerate chemotherapy regimens in this setting. Despite these concerns, however, there is growing evidence that advanced age alone is a poor predictor of the ability of an individual patient to tolerate and benefit from standard chemotherapy. The Elderly Lung Cancer Vinorelbine Italian Study Group was the first to publish a randomized trial that focuses specifically on elderly patients with advanced NSCLC.22 The Elderly Lung Cancer Vinorelbine Italian Study (ELVIS) trial has served as proof of principle that this subgroup can achieve survival and palliative benefit with systemic therapy. The investigational arm of the study was single-agent vinorelbine, which was compared with best supportive care. The trial was stopped early due to slow accrual. However, there was a significant survival difference that favored the treatment arm, with a 1-year survival of 32% versus 14% (P = 0.03). In addition, an analysis of QoL using the European Organization for Research and Treatment of Cancer (EORTC) questionnaires QLQ-C30 and CLQ-LC13 consistently favored the vinorelbine arm, although the difference was significant only for cognitive function.

Two subsequent trials have investigated the role of combination chemotherapy, both using nonplatinum regimens.11, 23 In the first trial Frasci et al. reported that patients age 70 years and older were randomized to receive gemcitabine plus vinorelbine versus vinorelbine alone.11 Patients who were randomized to the combination regimen had a 1-year survival rate of 30%, compared with 14% in the vinorelbine-alone arm (P < 0.01). A QoL analysis also favored patients in the combination arm. However, the survival of patients in the control arm clearly was worse than expected and was comparable to the no-treatment arm from the ELVIS trial. In addition, the toxicity of the combination arm was worse. In the Multicenter Italian Lung Cancer in the Elderly Study (MILES) trial, the same combination was compared with vinorelbine or gemcitabine alone.12 In that study, 698 patients were randomized to the 3 treatment arms. The response rate for the combination arm was 20%, compared with 18.5% for the vinorelbine arm and 17.3% for the gemcitabine arm: The respective 1-year survival rates were 31%, 41%, and 26%, and the differences were not significant when the combination arm was compared with either single-agent arm.

The role of cisplatin-based combination therapy in the management of elderly patients with lung carcinoma remains controversial. No randomized trials have addressed this issue. Although there have been reports that elderly patients can tolerate and benefit from cisplatin-based regimens, others have reported no benefit or a significant increase in associated toxicity.24–28 In a recent retrospective review of ECOG trial 5592, elderly patients (age 70 years and older) who were treated with cisplatin combined with either paclitaxel (24-hour and 3-hour infusions) or etoposide did as well as younger patients with respect to response rates and overall survival.27 In addition, except for increased leukopenia and neuropsychiatric toxicity, which were more common in elderly male patients, there were no significant differences based on age in the frequency of other, more common toxicities associated with these regimens. As in the current analysis, there were no differences in the changes of QoL outcomes over time.

Substituting carboplatin for cisplatin in combination with third-generation chemotherapy agents typically has resulted in equally effective regimens that are tolerated better.14, 15 Consequently, these regimens are used more often, particularly in the palliative setting. However, there have been even fewer reports regarding the efficacy of carboplatin-based combinations in elderly patients. Paclitaxel combined with carboplatin has become one of the most common front-line regimens used in the United States.17 Nakamura et al. recently reported that 3-hour infusions of paclitaxel as a single agent are safe and effective in elderly patients.29 In a retrospective review of data from the Fox Chase Cancer Center, carboplatin combined with paclitaxel, given either as a 24-hour infusion or as a 1-hour infusion, was tolerated by elderly patients as well as younger patients with comparable survival.30 A review of Southwest Oncology trials 9509 and 9308 included patients who were treated with carboplatin and paclitaxel as well as patients who were treated with cisplatin and vinorelbine and with cisplatin alone.31 The toxicity in those trials was comparable between elderly patients and younger patients (age 70 years and older vs. younger than age 70 years), but there was a trend toward shorter survival times in the elderly population. More elderly patients were able to complete treatment (6 cycles were planned) with carboplatin and paclitaxel (35%) than with cisplatin and vinorelbine (4%). In a comparison of patients younger than age 70 years with patients age 70 years and older, cessation of treatment due to toxicity with cisplatin and vinorelbine occurred significantly more often in elderly patients (29% vs. 46%, respectively; P = 0.003) compared with carboplatin and paclitaxel (11% vs. 16%, respectively; P = 0.37).

In the current analysis, fit elderly patients with Stage IIIB and IV NSCLC were able to do as well with this common carboplatin-based regimen as younger patients, without a significant increase in toxicity. Patients age 70 years and older were well represented in the current trial, accounting for 29% of the patient population. There was no significant difference in the response rate (28% vs. 20%; P = 0.28) between the two age groups. Although the time to progression was slightly longer in the elderly cohort (4.8 months vs. 3,0 months; P = 0.049), the median survival (7.1 months vs. 7.8 months) and the 1-year survival rate (34% vs. 30%) among patients age 70 years and older and among patients younger than age 70 years, respectively, were not significantly different (P = 0.65). In addition, although QoL declined for all patients, there was no difference in the rate of decline of QoL outcomes between the two age groups.

With two exceptions, the incidence of both hematologic and nonhematologic toxicity commonly associated with this regimen was comparable between the two age groups. There were higher rates of overall leukopenia as well as malaise and fatigue in elderly patients. However, neither difference was clinically relevant or had an adverse impact on the ability to deliver therapy. Both differences reflected an increase in low-grade toxicity, and when higher-grade toxicities were compared, the differences lost significance (Tables 4, 5). In the initial randomization, a greater proportion of elderly patients were randomized to Arm B (treatment until progression). In addition, there were more elderly patients with a limited PS (70–80%) compared with the younger subgroup (Table 1). Because toxicity was compared through the first four cycles, the greater proportion of elderly patients randomized to Arm B should not have affected the incidence of either toxicity. We analyzed the incidence of malaise and fatigue in the two PS subgroups and did not identify a greater incidence in patients with a limited PS (Grade 1–4 malaise and fatigue: 30% vs. 27% of patients with a PS of 70–80% vs. patients with a PS of 90–100%, respectively; P = 0.66). Therefore, this imbalance also is an unlikely explanation for the increased incidence of malaise and fatigue in the elderly population identified in the current analysis.

There were significant differences in the reasons for treatment discontinuation. Although a greater proportion of elderly patients were randomized to Arm B, these differences remained when Arm B was considered separately (Table 3). Patient or physician choice was the predominant reason that elderly patients discontinued therapy. The lower baseline PS within the elderly group may have contributed to a lower threshold for stopping treatment on the part of either the patient or the physician, despite the lack of serious toxicity. Alternatively, other factors that were unrelated to toxicity (i.e., transportation difficulties, lack of adequate support in the home, etc.) may have contributed to the decision by either the patient or the physician to stop treatment and therefore would not have been captured in this database. A smaller proportion of elderly patients went on to receive SL therapy. This suggests either that their PS did not recover adequately or that they or their treating physician remained less likely to consider additional treatment for a variety of reasons that could not be identified in the current analysis. We are in the process of developing a questionnaire to capture these data better in future trials of treatment for elderly patients with NSCLC.

In summary, the optimal treatment for elderly patients with advanced NSCLC remains uncertain. Although it has been shown that chemotherapy is effective in the management of elderly patients with NSCLC, the ideal regimen or strategy has not been identified. The ELVIS10 and MILES12 trials set a precedent for single agents in this setting. The most extensively studied chemotherapy agents include vinorelbine and gemcitabine. More recently, it also has been shown that the taxanes are safe in selected patients.29 The role of combination regimens, particularly platinum-based, continues to be an unresolved question. However, despite concerns that elderly patients are less likely to tolerate platinum-based combinations, we did not find that was the case with this commonly used front-line regimen. This finding also is consistent with the report of Langer et al.27 Therefore, elderly patients should continue to be included in clinical trials of platinum-based therapy and should not be excluded from these regimens based solely on age. An area that needs more research is the impact of certain comorbidities on the PS of these patients as well as the impact of certain comorbidities on the ability of elderly patients to tolerate either single-agent regimens or nonplatinum-based or platinum based combination regimens.