- Email: [email protected]
Evaluation of the optimal duration of chemotherapy in phase II trials for inoperable non-small-cell lung cancer (NSCLC)
Annals of Oncology 6: 993-997, 1995. O 1995 Kluwer Academic Publishers. Printed in the Netherlands.
Original article Evaluation of the optimal duration of chemotherapy in phase II trials for inoperable non-small-cell lung cancer (NSCLC) H. Larsen,1 J. B. Sorensen,1 A. L. Nielsen,2 P. Dombernowsky2 & H. H. Hansen1 1
Department of Oncology, Finsencenter, National University Hospital; Copenhagen, Denmark
2
Department of Oncology, Herlev Hospital, University of
Conclusions: The data indicate that patients with NSCLC included in phase II trials who have not yet achieved a Purpose: To determine the optimal duration of chemo- response to chemotherapy after 168 days on study have a low therapy in phase II trials for patients with non-small-cell lung likelihood (2%) of a subsequent response. Hence, treatment cessation at this point should be considered for noncancer (NSCLC). Patients and methods: The time from start of treatment responding patients. Continuation of treatment from day 84 until achievement of response according to WHO criteria to day 168 resulted in response in only 7 patients out of the total of 43 responses noted (16%). Thus, the toxic effects of was determined retrospectively in 8 phase II trials. Results: Response to chemotherapy consisting of 4 com- the chemotherapy in addition to the inconvenience of plete and 39 partial remissions was registered in 43 of 333 hospital visits renders it questionable whether it is worthpatients. The median time from treatment start to response while to continue treatment in patients with inoperable was 54 days. On day 84 on-study, 35 of the responding pa- non-small-cell lung cancer beyond day 84 in the absence of tients (81%) had achieved the response. Forty-three re- response. sponses (98%) had occurred by day 168 and only one patient (2%) accomplished a response after 168 days of treatment. Key words: chemotherapy, non-small-cell lung cancer, phase The responses had a median duration of 151 days (range 2 8 - II trials, response 1559 days). Summary
Introduction
Non-small-cell lung cancer (NSCLC) is one of the leading causes of cancer deaths worldwide. Only 25% of the patients have resectable disease at the time of diagnosis [1], but, in spite of complete resection, the 5-year survival rate for NSCLC in stages I, II, HIA, LOB and IV are only 46%-69%, 22%-40%, 8%-28%, 2%15%, 2%-15% and 0%-6%, respectively [2, 3], revealing recurrence of disease as a major problem in all stages of NSCLC. Chemotherapy for patients with non-resectable or recurrent NSCLC is experimental. Multicenter clinical trials with combination chemotherapy have demonstrated a very modest improvement in survival over that of best supportive care [4, 5] and these findings were statistically significant in a recent meta-analysis [6]. However, the survival advantage was small and in order to improve results in patients with NSCLC, further clinical investigations are necessary in the search for new agents with activity in NSCLC, as well as for evaluating new combination chemotherapy regimens. The conduct of such trials and the reporting of data are under continuous evaluation in an effort to obtain a maximum of reliable information with a ininimum of discomfort and inconvenience for the patients.
When optimizing the conduct of phase II trials it is important to determine the time the patients have to be on study with, on the one hand, a chance of achieving a response to treatment and, on the other hand, the potential risk of toxic effects from the drugs under investigation. Assessment of the necessary treatment duration is relatively straightforward in patients who have already achieved either a partial or a complete response, as these patients usually are treated until either progression, intolerable side effects, or, in the absence of such events, for a predetermined maximum treatment duration of, for instance, one year. Similarly, patients are taken off study if progression of the disease is detected, with or without a previous response. Very frequently no change in the disease parameters occurs during several treatment courses. Little is known about the length of time these patients have to continue on study to achieve the optimal cost-benefit with the highest possibility for response and the shortest exposure to potentially toxic drugs. This subject has been only briefly commented on in the literature. In a study by Sorensen et al. describing prognostic factors for response to treatment, the time from start of treatment until responses were achieved was also described [7]. The study included 218 evaluable patients with NSCLC, all with adenocarcinoma of
994 the lung (ACL) stage mA, mB, or IV. Of these, 53 patients achieved a response, 41 (77%) of which occurred within the first 84 days, while all but one were noted before 140 days of treatment had elapsed. In light of these results, the present investigation was initiated to assess the optimal treatment duration in phase II trials including patients with non-resectable NSCLC of all histologic types. Patients and methods
Results
The study is a retrospective evaluation including patients treated in 8 consecutive phase II trials at the departments of oncology at Herlev Hospital and the Finsencenter, National University Hospital. The chemotherapy trials were conducted during the period 19861993. A total of 8 consecutive phase II studies were analysed [8-15]. Doses and schedules are given in Table 1. The patients were evaluated at least every four weeks while on-study. All patients had non-resectable NSCLC with evaluable (unidimensionally measurable) or measurable (bidimensionally measurable) disease. Further inclusion criteria were age <70 years, ECOG performance status <2, absence of other previous or concomitant malignancy and informed consent. Patient pretreatment characteristics were recorded at the time of entry into the phase II trial protocols and collected from patient case reports. All data were obtained by the same observer in order to avoid interobserver variability. The responses were reviewed by two or more physicians and consensus was achieved. The evaluation of response was performed according to WHO criteria [16]. Complete response is the disappearance of all known disease determined by two observations not less than 4 weeks apart. Partial response is defined as a decrease of 50% or more in total tumour size of the rumor lesion in two observations not less than 4 weeks apart.
Table 1. Doses and schedules in the 8 phase II trials included in the study. Regimen
Dose (mg/m 2 )
Route of Schedule administration
Ref.
TCNU (single dose)
130
p.o.
once q 4 wks
[4]
TCNU (split-dose)
40
p.o.
daily in 3 days q 4 wks
[5]
110 3
p.o. i.v.
once q 4 wks weekly x 8, then biweekly
[31
VM-26
80
i.v.
daily for 5 days q 3 wks
[61
IDX
80
i.v.
once q 3 wks
[7]
1000
i.v.
weekly x 3 q 4 wks
[11]
Gemcitabine (twice weekly)
90
i.v.
twice weekly 3 days interval
[10]
Gemcitabine + VDS
1000 3
weekly x 3 q 4 wks q 2 wks
[141
TCNU+VDS
Gemcitabine (weekly)
In order to optimize the cost-benefit in trials, i.e., the oplimal time on study, the endpoint was the time from start of treatmcnl to the first observation of a partial or complete response. The first observation of a more than 50% tumor reduction was used, in contrasl to the WHO criteria in which the reduction must have lasted for a minimum of 4 weeks. This method was applied in order to detect the earliest point on study at which the treatment could be stopped in case of no change. The numbers of delivered cycles of chemotherapy are not analysed because of the varying treatment schedules.
i.v. i.v.
Abbreviations: p.o. - peroral; i.v. - intravenous; q - every; wks weeks; TCNU — l-(2-chloroethyl)-3-[2-(dimethylaminosulfonyl)etyl]-l-nitrosurea; VDS - vindesine; IDX — iododoxorubicine; VM-26 - teniposide.
Data were collected from a total of 336 patients, of whom 333 were evaluable for response and included in the analyses. At the end of the observation time 316 patients had died, 17 were still alive and 3 were still on study. Their pretreatment characteristics are shown in Table 2. The distribution of histologic groups reveals a high fraction of adenocarcinoma (69%). With respect to gender, the male/female ratio is equal, reflecting the
Table 2. Pretreatment characteristics of the 333 patients included in the 8 phase II studies evaluated. Characteristics Age Median Range Gender Male Female Histology Squamous cell carcinoma Adenocarcinoma Large-cell carcinoma Adeno-squamous carcinoma Poorly differentiated carcinoma Stage" T i TT 1 + 11 I1IA niR IV Performance status 0 1 2 Prior surgery None Complete resectionb Subsegmental resection Lobectomy Pneumonectomy Incomplete resection Explorative Subsegmental resection Lobectomy Pneumonectomy
n
(%) 58.4 33.6-73.5
169 164
(51) (49)
54 231 38 5
(16) (69) (11) (2) (2)
11 85 94 143
(3) (26) (28) (43)
69 196 68
(20) (60) (20)
279 6
(84) (2)
• 5
1 4 1 48
(0) (1) (0) (14)
37 5 4 2
(11) (2) (1) (1)
* Patients in stage I and II were physically unfit for surgery. b The patients considered radically resected had recurrence of disease when entering the phase II trial.
995 high fraction of adenocarcinoma. A total of 279 patients had had no prior surgery, 6 had relapsed following apparently complete resection and 48 patients had had incomplete resection or explorative thoractomy only. The median treatment duration for the 333 patients was 91 days (range 2-662 days). The treatment results are outlined in greater detail in Table 3, showing the number of evaluable patients included in the individual studies together with the number of partial and complete responses. Forty-three responses were seen, 39 of them partial and four complete. The time from start of treatment to initial response at specific landmark times is shown in Table 4. The median time from start of treatment to response was 54 days (range 14-260 days), demonstrated in greater detail in Fig. 1. Forty-two of the responses (98%) had occurred by day 168. The possibility of subsequent responses among the non-responding patients still on study at this point was only 2%. Time to response is rather similar in the respective studies despite the use
112 *U
M3
140
MB <2 47
D1
42 40
Fig. 1. Cumulative number of responses. Based on time from start of treatment to first sign of response.
of different drugs and schedules and apparently independently of whether the therapy comprised one or two drugs, except in the study evaluating lododoxorubicine where the only response noted occurred 260 days after start of treatment. When analysing the time to response among the histologic subtypes no differences were observed.
Table 3. Number of patients evaluable for response included in the eight phase II trials and treatment results (median survival and number of partial and complete responders. Regimen
No. of pts
Response
Median
Range
PR:n
(%)
CRVn
(%)
58 39 47 43 23 32 58 33
139 168 149 146 119 166 163 140
5-1619 18-1587 3-679 0-886 15-1039 2-784 5-784 14-482
6 3 6 1 2 5 11 9
(9.7) (7.7) (12.8) (2.2) (8.3) (15.6) (19.0) (27.3)
2 1 0 0 0 0 0 1
(3.4) (2.6)
333
150
0-1619
43
(12.9)
4
(1.2)
TCNU (single dose) TCNU (split dose) TCNU + VDS IDX VM 26 Gemcitabine (weekly) Gemcitabine (twice weekly) Gemcitabine + VDS Total
Survival (days)
(2.7)
' The four CR were initially registered as PR and are therefore present in both columns. Abbreviations: PR - partial response; CR - complete response; TCNU - l-(2-chloroethyl)-3-[2-(dimethylaminosulfonyl)ethyl]-l-nitrosurea; VDS — vindesine; IDX - iododoxorubicine; VM-26 - teniposide.
Table 2. Cumulative number of responses achieved at different landmark times in the eight trials included in the study. Regimen
TCNU (single dose) TCNU (split-dose) TCNU + VDS VM-26 4-IDX Gemcitabine (weekly) Gemcitabine (twice weekly) Gemcitabine + VDS Total
Days on study 28
56
84
112
140
168
196
2 2 3 0 0 0 2 2
5 2 5 1 0 4 7 5
6 3 6 2 0 4 9 5
NFR NFR NFR NFR 0 4 9 7
0 5 10 9
0 NFR 11 NFR
0
11
29
35
37
41
42
224
252
280
>280
NFR
NFR
42
42
42
43
NFR
Abbreviations: NFR — no further responses recorded; TCNU - l-{2-chloroethyl)-3-[2-(dimethylaminosulfonyl)ethyl]-l-nitrosurea; VDS — vindesine; IDX - iododoxorubicine; VM-26 = teniposide.
996 The number of patients still on study at various landmark times, and thus with either a response or with the possibility of achieving a subsequent response are shown in Table 5. The figures are divided into one- and two-drug schedules and demonstrates similar fractions of patients still on study at different landmark times. After 84 days of treatment a total of 183 patients (55%) are still on study, while after day 168 the corresponding number has decreased to 79 patients (24%), and after day 252 only 40 patients (12%) are still on study. The fraction of responding patients among those still on study increases with increasing treatment time. The ratio between the number of subsequent responses to be achieved by the number of patients without response still on study shows a decreasing possibility for response at the same landmark times (Table 6). A total of 149 patients (81%) are without response after 84 days of treatment, and have a 5% possibility of achieving a subsequent response. After 168 days on study, 53 patients (67%) had not achieved a response, and had a 2% change of achieving a response. Overall, the median treatment time for non-responding patients was 84 days (range 0-662 days). The median treatment time for responding patients was 210 days (range 76-567). Sixty-six non-responding patients (19%) were treated for more than 140 days, and 30 non-responding patients (9%) for more than 208 days. The median survival for all of the patients was 150 days (range 0-1619), while the median survivals were 361 days (range 76-1619 days) for responding patients and 132 days (range 0-1039 days) for non-responding patients. Discussion The optimal duration of chemotherapy was first discussed in a study by Serensen et al. [7] evaluating 218 patients, all with ACL. The present study included 333 patients of all major histologic types of NSCLC. After 84 days of treatment, 41 patients (77%) in the study by Sorensen et al. had achieved a response, while in the present investigation 35 responses (81%) have been noted. Only one patient (2%) in the Sorensen study achieved a response after 20 weeks (140 days) of treatment, while in the present study 2 patients (5%) achieved a response after day 140. The data from our
Table 6 Number of responding patients at different landmark times and concurrent number of patients still on study (all 8 phase U trials summarized). Time (days)
Total no. pts. still on study n
0 28 56 84 112 140 168 196 224 252 280 >280
333 292 230 183 130 96 79 59 47 40 37 29
%
Status of patients still on trial With wim
wimoui
response
response
n
%
n
11 29 34 31 30 26 24 20 16 16 12
(4) (13) (19) (24) (31) (33) (41) (43) (40) (43) (41)
334 (100) 281 (96) 201 (87) 149 (81) 99 (76) 66 (69) 53 (67) 35 (59) 27 (57) 24 (60) 21 (57) 17 (59)
(100) (88) (69) (55) (39) (29) (23) (18) (14) (12) (11) (9)
Cumu- Chance lative for subno. of sequent reresponses sponse (%)• n
% 11 29 35 37 41 42 42 42 42 43 NFR
13 11 7 5 6 3 2 3 4 4 0 0
Abbreviations: NFR - no further responses recorded. ' The ratio between remaining number of responses to be achieved divided by the number of patients still on study without response.
study is in accord with these observations, indicating that the data on ACL are in accord with the data on NSCLC including all major histologic types. However, the majority of patients in the present study also had ACL, while other histologic types of NSCLC accounted for only one-third of the cases. A former study by Sorensen et al. [9] employed one-, three- and four-drug combinations, while the present studies have used one- and two-drug combinations. The number of patients evaluated in these studies are 324 receiving one-drug treatment, 80 receiving two drugs, 74 receiving three drugs and 73 patients receiving four drugs. There appear to be no differences in the time to first observation of response in these studies. The present study shows that the time to response is relatively short (median 54 days). Thus, it is possible within a few months of treatment to define patients with a low likelihood of subsequent response. Platinum-containing chemotherapy was not used in any of the phase n studies included in the evaluation. Future
Table 5. Number of patients (and percentages) still on study at different landmark times divided according to treatment with one and two drug schedules, i.e., patients with response or with possibility of subsequent response. Regimen
One-drug schedules Two-drug schedules Total
Days on study 0
28
56
84
253(100)
222(88)
173(68)
137(54)
80(100)
70(88)
57(71)
333(100)
292(88)
230(69)
112
140 168 n (%)
196
224
252
280
>280
97(38)
72(28)
61(24)
47(19)
37(15)
32(13)
29(12)
21(8)
46(58)
33(41)
24(30)
18(23)
12(15)
10(13)
8(10)
8(10)
183(55)
130(39)
96(29)
79(24)
59(18)
47(14)
40(12)
37(11)
8(10) 29(9)
997 studies will have to test whether the same conclusions may be drawn about platinum-containing chemotherapy regimens. It is difficult to know whether the non-responding patients included in phase II trials obtained any advantage from the treatment. It was previously demonstrated that treatment can diminish disease-related symptoms such as cough, haemoptysis, bone pain, malaise and weight loss in approximately 70% of patients treated with combination chemotherapy for advanced NSCLC [17]. Improvement is usually related to objective response but the fraction of patients with improved symptoms usually exceeds the response rate. The same beneficial effect may to some extent occur in phase II studies, although it has not been definitively proven in such trials. The potential reduction in quality of life due to toxic effects of chemotherapy is difficult to assess, but the cumulative risk for toxicity and thereby a possible detrimental effect on the quality of life increases with increasing length of the treatment. This aspect remains controversial and must be balanced against the fact that new agents active in the treatment of NSCLC are urgently needed. Accordingly, the treatment time for patients who do not obtain an objective benefit from the chemotherapy should be as short as possible in order for the optimal cost-benefit from the treatment to be achieved. Based on the present analysis, we conclude that patients with NSCLC who have not responded to treatment at the latest by day 168 should discontinue the chemotherapy. In addition, it is questionable whether it is worthwhile to continue treatment longer than day 84 if response has not been achieved by then. Such an approach may call for consideration of the statistics on which the design of phase n studies is based. The calculated number of patients to be included in the studies is usually based on the selected power of the study to discover a certain response rate. In these calculations it is assumed that the chances for response are the same for all patients included, i.e., all patients receive chemotherapy until the appearance of an event such as unacceptable toxic effects, progression of disease or death. When choosing to continue treatment until only 90% have had the maximum chance for response, it may be necessary to include a larger number of patients. Studies conducted in this way may be more cost-effective, both in view of the quality of life for the participating patients and the costs associated with the treatment. References 1. Freise G, Gabler A, Liebig S. Bronchial carcinoma and longterm survival. Thorax 1978; 33: 228-34.
2. Mountain CF. Prognostic implication of the International Staging System for lung cancer. Semin Oncol 1988; 15: 23645. 3. Naruke T, Goya T, Tsuchiya R, Suemasu K. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988; 96: 440-7. 4. Rapp E, Pater JL, Willan A et al. Chemotherapy can prolong survival in patients with advanced non-small-cell lung cancer report of a Canadian multicenter randomized trial. J Clin Oncol 1988; 6:633-41. 5. Cormier Y, Bergerson D, LaForge J. Benefits of polychemotherapy in advanced non-small-cell bronchogenic carcinoma. Cancer 1982; 50: 845-9. 6. Souquet PJ, Chauvin F, Boissel JP et al. Polychemotherapy in advanced non-small-cell lung cancer A metaanalysis. Lancet 1993; 342:19-21. 7. Sorensen JB, Badsberg JH, Hansen HH. Response to cytostatic treatment in inoperable adenocarcinoma of the lung: Critical implications. Br J Cancer 1989; 60: 389-93. 8. Vibe-Petersen J, Bach F, Pedersen AG et al. A phase II trial of TCNU in patients with squamous, adeno and large-cell carcinoma of the lung. Eur J Cancer Clin Oncol 1990; 25:1881-5. 9. Serensen JB, Hansen HH, Dombernowsky P et al. chemotherapy for adenocarcinoma of the lung (WHO 3): A randomized study of vindesine versus lomustine, cyclophosphamide and methotrexate versus all four drugs. J Clin Oncol 1987; 5: 116977. 10. Bach F, Vibe-Petersen J, Serensen JB et al. Phase II trial of TCNU and vindesine in patients with adenocarcinomaof the lung. Eur J Cancer 1990; 26: 545. 11. Sarensen JB, Stenbygaard L, Drivsholm L et al. Phase U study of 4-iodo-4-deoxydoxorubicin in non-resectable non-small-cell lung cancer. Cancer Chemother Pharmacol 1993; 32: 399402. 12. Anderson H, Lund B, Bach F et al. Single-agent activity of weekly gemcitabine in advanced non-small-cell lung cancer: A phase H study. J Clin Oncol 1994; 12:1821-6. 13. Lund B, Ryberg M, Meidal Petersen P et al. Phase II study of gemcitabine (2,2-difluorodeoxycytidine) given as a twice-weekly schedule to previously untreated patients with non-small-cell lung cancer. Ann Oncol 1994; 5:852-3. 14. Serensen JB, Bach F, Dombernowsky P, Hansen HH. Phase H study of teniposide in adenocarcinoma of the lung. Cancer Chemother Pharmacol 1991; 27:487-9. 15. S0rensen JB, Nielsen AL, Krarup M et al. Phase II study of gemcitabine + vindesine in patients with previously untreated, inoperable non-small-cell lung cancer (NSCLC). Lung Cancer 1994; 11 (Suppl 1)P:116. 16. Anonymous. WHO Handbook for Reporting Results of Cancer Treatment. WHO Offset Publication 1979; 48. 17. Osoba D, Rusthoven JJ, Turnbull KA et al. Combination chemotherapy with bleomycin, etoposide, and cisplatin in metastatic non-small-cell lung cancer. J Clin Oncol 1985; 3: 1478-85. Received 20 July 1995; accepted 4 October 1995.
Correspondence to: Henrik Larsen, M.D. Department of Oncology 5072 National University Hospital/Rigshospitalet 9 Blegdamsvej DK-2100 Copenhagen, Denmark
Original article Evaluation of the optimal duration of chemotherapy in phase II trials for inoperable non-small-cell lung cancer (NSCLC) H. Larsen,1 J. B. Sorensen,1 A. L. Nielsen,2 P. Dombernowsky2 & H. H. Hansen1 1
Department of Oncology, Finsencenter, National University Hospital; Copenhagen, Denmark
2
Department of Oncology, Herlev Hospital, University of
Conclusions: The data indicate that patients with NSCLC included in phase II trials who have not yet achieved a Purpose: To determine the optimal duration of chemo- response to chemotherapy after 168 days on study have a low therapy in phase II trials for patients with non-small-cell lung likelihood (2%) of a subsequent response. Hence, treatment cessation at this point should be considered for noncancer (NSCLC). Patients and methods: The time from start of treatment responding patients. Continuation of treatment from day 84 until achievement of response according to WHO criteria to day 168 resulted in response in only 7 patients out of the total of 43 responses noted (16%). Thus, the toxic effects of was determined retrospectively in 8 phase II trials. Results: Response to chemotherapy consisting of 4 com- the chemotherapy in addition to the inconvenience of plete and 39 partial remissions was registered in 43 of 333 hospital visits renders it questionable whether it is worthpatients. The median time from treatment start to response while to continue treatment in patients with inoperable was 54 days. On day 84 on-study, 35 of the responding pa- non-small-cell lung cancer beyond day 84 in the absence of tients (81%) had achieved the response. Forty-three re- response. sponses (98%) had occurred by day 168 and only one patient (2%) accomplished a response after 168 days of treatment. Key words: chemotherapy, non-small-cell lung cancer, phase The responses had a median duration of 151 days (range 2 8 - II trials, response 1559 days). Summary
Introduction
Non-small-cell lung cancer (NSCLC) is one of the leading causes of cancer deaths worldwide. Only 25% of the patients have resectable disease at the time of diagnosis [1], but, in spite of complete resection, the 5-year survival rate for NSCLC in stages I, II, HIA, LOB and IV are only 46%-69%, 22%-40%, 8%-28%, 2%15%, 2%-15% and 0%-6%, respectively [2, 3], revealing recurrence of disease as a major problem in all stages of NSCLC. Chemotherapy for patients with non-resectable or recurrent NSCLC is experimental. Multicenter clinical trials with combination chemotherapy have demonstrated a very modest improvement in survival over that of best supportive care [4, 5] and these findings were statistically significant in a recent meta-analysis [6]. However, the survival advantage was small and in order to improve results in patients with NSCLC, further clinical investigations are necessary in the search for new agents with activity in NSCLC, as well as for evaluating new combination chemotherapy regimens. The conduct of such trials and the reporting of data are under continuous evaluation in an effort to obtain a maximum of reliable information with a ininimum of discomfort and inconvenience for the patients.
When optimizing the conduct of phase II trials it is important to determine the time the patients have to be on study with, on the one hand, a chance of achieving a response to treatment and, on the other hand, the potential risk of toxic effects from the drugs under investigation. Assessment of the necessary treatment duration is relatively straightforward in patients who have already achieved either a partial or a complete response, as these patients usually are treated until either progression, intolerable side effects, or, in the absence of such events, for a predetermined maximum treatment duration of, for instance, one year. Similarly, patients are taken off study if progression of the disease is detected, with or without a previous response. Very frequently no change in the disease parameters occurs during several treatment courses. Little is known about the length of time these patients have to continue on study to achieve the optimal cost-benefit with the highest possibility for response and the shortest exposure to potentially toxic drugs. This subject has been only briefly commented on in the literature. In a study by Sorensen et al. describing prognostic factors for response to treatment, the time from start of treatment until responses were achieved was also described [7]. The study included 218 evaluable patients with NSCLC, all with adenocarcinoma of
994 the lung (ACL) stage mA, mB, or IV. Of these, 53 patients achieved a response, 41 (77%) of which occurred within the first 84 days, while all but one were noted before 140 days of treatment had elapsed. In light of these results, the present investigation was initiated to assess the optimal treatment duration in phase II trials including patients with non-resectable NSCLC of all histologic types. Patients and methods
Results
The study is a retrospective evaluation including patients treated in 8 consecutive phase II trials at the departments of oncology at Herlev Hospital and the Finsencenter, National University Hospital. The chemotherapy trials were conducted during the period 19861993. A total of 8 consecutive phase II studies were analysed [8-15]. Doses and schedules are given in Table 1. The patients were evaluated at least every four weeks while on-study. All patients had non-resectable NSCLC with evaluable (unidimensionally measurable) or measurable (bidimensionally measurable) disease. Further inclusion criteria were age <70 years, ECOG performance status <2, absence of other previous or concomitant malignancy and informed consent. Patient pretreatment characteristics were recorded at the time of entry into the phase II trial protocols and collected from patient case reports. All data were obtained by the same observer in order to avoid interobserver variability. The responses were reviewed by two or more physicians and consensus was achieved. The evaluation of response was performed according to WHO criteria [16]. Complete response is the disappearance of all known disease determined by two observations not less than 4 weeks apart. Partial response is defined as a decrease of 50% or more in total tumour size of the rumor lesion in two observations not less than 4 weeks apart.
Table 1. Doses and schedules in the 8 phase II trials included in the study. Regimen
Dose (mg/m 2 )
Route of Schedule administration
Ref.
TCNU (single dose)
130
p.o.
once q 4 wks
[4]
TCNU (split-dose)
40
p.o.
daily in 3 days q 4 wks
[5]
110 3
p.o. i.v.
once q 4 wks weekly x 8, then biweekly
[31
VM-26
80
i.v.
daily for 5 days q 3 wks
[61
IDX
80
i.v.
once q 3 wks
[7]
1000
i.v.
weekly x 3 q 4 wks
[11]
Gemcitabine (twice weekly)
90
i.v.
twice weekly 3 days interval
[10]
Gemcitabine + VDS
1000 3
weekly x 3 q 4 wks q 2 wks
[141
TCNU+VDS
Gemcitabine (weekly)
In order to optimize the cost-benefit in trials, i.e., the oplimal time on study, the endpoint was the time from start of treatmcnl to the first observation of a partial or complete response. The first observation of a more than 50% tumor reduction was used, in contrasl to the WHO criteria in which the reduction must have lasted for a minimum of 4 weeks. This method was applied in order to detect the earliest point on study at which the treatment could be stopped in case of no change. The numbers of delivered cycles of chemotherapy are not analysed because of the varying treatment schedules.
i.v. i.v.
Abbreviations: p.o. - peroral; i.v. - intravenous; q - every; wks weeks; TCNU — l-(2-chloroethyl)-3-[2-(dimethylaminosulfonyl)etyl]-l-nitrosurea; VDS - vindesine; IDX — iododoxorubicine; VM-26 - teniposide.
Data were collected from a total of 336 patients, of whom 333 were evaluable for response and included in the analyses. At the end of the observation time 316 patients had died, 17 were still alive and 3 were still on study. Their pretreatment characteristics are shown in Table 2. The distribution of histologic groups reveals a high fraction of adenocarcinoma (69%). With respect to gender, the male/female ratio is equal, reflecting the
Table 2. Pretreatment characteristics of the 333 patients included in the 8 phase II studies evaluated. Characteristics Age Median Range Gender Male Female Histology Squamous cell carcinoma Adenocarcinoma Large-cell carcinoma Adeno-squamous carcinoma Poorly differentiated carcinoma Stage" T i TT 1 + 11 I1IA niR IV Performance status 0 1 2 Prior surgery None Complete resectionb Subsegmental resection Lobectomy Pneumonectomy Incomplete resection Explorative Subsegmental resection Lobectomy Pneumonectomy
n
(%) 58.4 33.6-73.5
169 164
(51) (49)
54 231 38 5
(16) (69) (11) (2) (2)
11 85 94 143
(3) (26) (28) (43)
69 196 68
(20) (60) (20)
279 6
(84) (2)
• 5
1 4 1 48
(0) (1) (0) (14)
37 5 4 2
(11) (2) (1) (1)
* Patients in stage I and II were physically unfit for surgery. b The patients considered radically resected had recurrence of disease when entering the phase II trial.
995 high fraction of adenocarcinoma. A total of 279 patients had had no prior surgery, 6 had relapsed following apparently complete resection and 48 patients had had incomplete resection or explorative thoractomy only. The median treatment duration for the 333 patients was 91 days (range 2-662 days). The treatment results are outlined in greater detail in Table 3, showing the number of evaluable patients included in the individual studies together with the number of partial and complete responses. Forty-three responses were seen, 39 of them partial and four complete. The time from start of treatment to initial response at specific landmark times is shown in Table 4. The median time from start of treatment to response was 54 days (range 14-260 days), demonstrated in greater detail in Fig. 1. Forty-two of the responses (98%) had occurred by day 168. The possibility of subsequent responses among the non-responding patients still on study at this point was only 2%. Time to response is rather similar in the respective studies despite the use
112 *U
M3
140
MB <2 47
D1
42 40
Fig. 1. Cumulative number of responses. Based on time from start of treatment to first sign of response.
of different drugs and schedules and apparently independently of whether the therapy comprised one or two drugs, except in the study evaluating lododoxorubicine where the only response noted occurred 260 days after start of treatment. When analysing the time to response among the histologic subtypes no differences were observed.
Table 3. Number of patients evaluable for response included in the eight phase II trials and treatment results (median survival and number of partial and complete responders. Regimen
No. of pts
Response
Median
Range
PR:n
(%)
CRVn
(%)
58 39 47 43 23 32 58 33
139 168 149 146 119 166 163 140
5-1619 18-1587 3-679 0-886 15-1039 2-784 5-784 14-482
6 3 6 1 2 5 11 9
(9.7) (7.7) (12.8) (2.2) (8.3) (15.6) (19.0) (27.3)
2 1 0 0 0 0 0 1
(3.4) (2.6)
333
150
0-1619
43
(12.9)
4
(1.2)
TCNU (single dose) TCNU (split dose) TCNU + VDS IDX VM 26 Gemcitabine (weekly) Gemcitabine (twice weekly) Gemcitabine + VDS Total
Survival (days)
(2.7)
' The four CR were initially registered as PR and are therefore present in both columns. Abbreviations: PR - partial response; CR - complete response; TCNU - l-(2-chloroethyl)-3-[2-(dimethylaminosulfonyl)ethyl]-l-nitrosurea; VDS — vindesine; IDX - iododoxorubicine; VM-26 - teniposide.
Table 2. Cumulative number of responses achieved at different landmark times in the eight trials included in the study. Regimen
TCNU (single dose) TCNU (split-dose) TCNU + VDS VM-26 4-IDX Gemcitabine (weekly) Gemcitabine (twice weekly) Gemcitabine + VDS Total
Days on study 28
56
84
112
140
168
196
2 2 3 0 0 0 2 2
5 2 5 1 0 4 7 5
6 3 6 2 0 4 9 5
NFR NFR NFR NFR 0 4 9 7
0 5 10 9
0 NFR 11 NFR
0
11
29
35
37
41
42
224
252
280
>280
NFR
NFR
42
42
42
43
NFR
Abbreviations: NFR — no further responses recorded; TCNU - l-{2-chloroethyl)-3-[2-(dimethylaminosulfonyl)ethyl]-l-nitrosurea; VDS — vindesine; IDX - iododoxorubicine; VM-26 = teniposide.
996 The number of patients still on study at various landmark times, and thus with either a response or with the possibility of achieving a subsequent response are shown in Table 5. The figures are divided into one- and two-drug schedules and demonstrates similar fractions of patients still on study at different landmark times. After 84 days of treatment a total of 183 patients (55%) are still on study, while after day 168 the corresponding number has decreased to 79 patients (24%), and after day 252 only 40 patients (12%) are still on study. The fraction of responding patients among those still on study increases with increasing treatment time. The ratio between the number of subsequent responses to be achieved by the number of patients without response still on study shows a decreasing possibility for response at the same landmark times (Table 6). A total of 149 patients (81%) are without response after 84 days of treatment, and have a 5% possibility of achieving a subsequent response. After 168 days on study, 53 patients (67%) had not achieved a response, and had a 2% change of achieving a response. Overall, the median treatment time for non-responding patients was 84 days (range 0-662 days). The median treatment time for responding patients was 210 days (range 76-567). Sixty-six non-responding patients (19%) were treated for more than 140 days, and 30 non-responding patients (9%) for more than 208 days. The median survival for all of the patients was 150 days (range 0-1619), while the median survivals were 361 days (range 76-1619 days) for responding patients and 132 days (range 0-1039 days) for non-responding patients. Discussion The optimal duration of chemotherapy was first discussed in a study by Serensen et al. [7] evaluating 218 patients, all with ACL. The present study included 333 patients of all major histologic types of NSCLC. After 84 days of treatment, 41 patients (77%) in the study by Sorensen et al. had achieved a response, while in the present investigation 35 responses (81%) have been noted. Only one patient (2%) in the Sorensen study achieved a response after 20 weeks (140 days) of treatment, while in the present study 2 patients (5%) achieved a response after day 140. The data from our
Table 6 Number of responding patients at different landmark times and concurrent number of patients still on study (all 8 phase U trials summarized). Time (days)
Total no. pts. still on study n
0 28 56 84 112 140 168 196 224 252 280 >280
333 292 230 183 130 96 79 59 47 40 37 29
%
Status of patients still on trial With wim
wimoui
response
response
n
%
n
11 29 34 31 30 26 24 20 16 16 12
(4) (13) (19) (24) (31) (33) (41) (43) (40) (43) (41)
334 (100) 281 (96) 201 (87) 149 (81) 99 (76) 66 (69) 53 (67) 35 (59) 27 (57) 24 (60) 21 (57) 17 (59)
(100) (88) (69) (55) (39) (29) (23) (18) (14) (12) (11) (9)
Cumu- Chance lative for subno. of sequent reresponses sponse (%)• n
% 11 29 35 37 41 42 42 42 42 43 NFR
13 11 7 5 6 3 2 3 4 4 0 0
Abbreviations: NFR - no further responses recorded. ' The ratio between remaining number of responses to be achieved divided by the number of patients still on study without response.
study is in accord with these observations, indicating that the data on ACL are in accord with the data on NSCLC including all major histologic types. However, the majority of patients in the present study also had ACL, while other histologic types of NSCLC accounted for only one-third of the cases. A former study by Sorensen et al. [9] employed one-, three- and four-drug combinations, while the present studies have used one- and two-drug combinations. The number of patients evaluated in these studies are 324 receiving one-drug treatment, 80 receiving two drugs, 74 receiving three drugs and 73 patients receiving four drugs. There appear to be no differences in the time to first observation of response in these studies. The present study shows that the time to response is relatively short (median 54 days). Thus, it is possible within a few months of treatment to define patients with a low likelihood of subsequent response. Platinum-containing chemotherapy was not used in any of the phase n studies included in the evaluation. Future
Table 5. Number of patients (and percentages) still on study at different landmark times divided according to treatment with one and two drug schedules, i.e., patients with response or with possibility of subsequent response. Regimen
One-drug schedules Two-drug schedules Total
Days on study 0
28
56
84
253(100)
222(88)
173(68)
137(54)
80(100)
70(88)
57(71)
333(100)
292(88)
230(69)
112
140 168 n (%)
196
224
252
280
>280
97(38)
72(28)
61(24)
47(19)
37(15)
32(13)
29(12)
21(8)
46(58)
33(41)
24(30)
18(23)
12(15)
10(13)
8(10)
8(10)
183(55)
130(39)
96(29)
79(24)
59(18)
47(14)
40(12)
37(11)
8(10) 29(9)
997 studies will have to test whether the same conclusions may be drawn about platinum-containing chemotherapy regimens. It is difficult to know whether the non-responding patients included in phase II trials obtained any advantage from the treatment. It was previously demonstrated that treatment can diminish disease-related symptoms such as cough, haemoptysis, bone pain, malaise and weight loss in approximately 70% of patients treated with combination chemotherapy for advanced NSCLC [17]. Improvement is usually related to objective response but the fraction of patients with improved symptoms usually exceeds the response rate. The same beneficial effect may to some extent occur in phase II studies, although it has not been definitively proven in such trials. The potential reduction in quality of life due to toxic effects of chemotherapy is difficult to assess, but the cumulative risk for toxicity and thereby a possible detrimental effect on the quality of life increases with increasing length of the treatment. This aspect remains controversial and must be balanced against the fact that new agents active in the treatment of NSCLC are urgently needed. Accordingly, the treatment time for patients who do not obtain an objective benefit from the chemotherapy should be as short as possible in order for the optimal cost-benefit from the treatment to be achieved. Based on the present analysis, we conclude that patients with NSCLC who have not responded to treatment at the latest by day 168 should discontinue the chemotherapy. In addition, it is questionable whether it is worthwhile to continue treatment longer than day 84 if response has not been achieved by then. Such an approach may call for consideration of the statistics on which the design of phase n studies is based. The calculated number of patients to be included in the studies is usually based on the selected power of the study to discover a certain response rate. In these calculations it is assumed that the chances for response are the same for all patients included, i.e., all patients receive chemotherapy until the appearance of an event such as unacceptable toxic effects, progression of disease or death. When choosing to continue treatment until only 90% have had the maximum chance for response, it may be necessary to include a larger number of patients. Studies conducted in this way may be more cost-effective, both in view of the quality of life for the participating patients and the costs associated with the treatment. References 1. Freise G, Gabler A, Liebig S. Bronchial carcinoma and longterm survival. Thorax 1978; 33: 228-34.
2. Mountain CF. Prognostic implication of the International Staging System for lung cancer. Semin Oncol 1988; 15: 23645. 3. Naruke T, Goya T, Tsuchiya R, Suemasu K. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988; 96: 440-7. 4. Rapp E, Pater JL, Willan A et al. Chemotherapy can prolong survival in patients with advanced non-small-cell lung cancer report of a Canadian multicenter randomized trial. J Clin Oncol 1988; 6:633-41. 5. Cormier Y, Bergerson D, LaForge J. Benefits of polychemotherapy in advanced non-small-cell bronchogenic carcinoma. Cancer 1982; 50: 845-9. 6. Souquet PJ, Chauvin F, Boissel JP et al. Polychemotherapy in advanced non-small-cell lung cancer A metaanalysis. Lancet 1993; 342:19-21. 7. Sorensen JB, Badsberg JH, Hansen HH. Response to cytostatic treatment in inoperable adenocarcinoma of the lung: Critical implications. Br J Cancer 1989; 60: 389-93. 8. Vibe-Petersen J, Bach F, Pedersen AG et al. A phase II trial of TCNU in patients with squamous, adeno and large-cell carcinoma of the lung. Eur J Cancer Clin Oncol 1990; 25:1881-5. 9. Serensen JB, Hansen HH, Dombernowsky P et al. chemotherapy for adenocarcinoma of the lung (WHO 3): A randomized study of vindesine versus lomustine, cyclophosphamide and methotrexate versus all four drugs. J Clin Oncol 1987; 5: 116977. 10. Bach F, Vibe-Petersen J, Serensen JB et al. Phase II trial of TCNU and vindesine in patients with adenocarcinomaof the lung. Eur J Cancer 1990; 26: 545. 11. Sarensen JB, Stenbygaard L, Drivsholm L et al. Phase U study of 4-iodo-4-deoxydoxorubicin in non-resectable non-small-cell lung cancer. Cancer Chemother Pharmacol 1993; 32: 399402. 12. Anderson H, Lund B, Bach F et al. Single-agent activity of weekly gemcitabine in advanced non-small-cell lung cancer: A phase H study. J Clin Oncol 1994; 12:1821-6. 13. Lund B, Ryberg M, Meidal Petersen P et al. Phase II study of gemcitabine (2,2-difluorodeoxycytidine) given as a twice-weekly schedule to previously untreated patients with non-small-cell lung cancer. Ann Oncol 1994; 5:852-3. 14. Serensen JB, Bach F, Dombernowsky P, Hansen HH. Phase H study of teniposide in adenocarcinoma of the lung. Cancer Chemother Pharmacol 1991; 27:487-9. 15. S0rensen JB, Nielsen AL, Krarup M et al. Phase II study of gemcitabine + vindesine in patients with previously untreated, inoperable non-small-cell lung cancer (NSCLC). Lung Cancer 1994; 11 (Suppl 1)P:116. 16. Anonymous. WHO Handbook for Reporting Results of Cancer Treatment. WHO Offset Publication 1979; 48. 17. Osoba D, Rusthoven JJ, Turnbull KA et al. Combination chemotherapy with bleomycin, etoposide, and cisplatin in metastatic non-small-cell lung cancer. J Clin Oncol 1985; 3: 1478-85. Received 20 July 1995; accepted 4 October 1995.
Correspondence to: Henrik Larsen, M.D. Department of Oncology 5072 National University Hospital/Rigshospitalet 9 Blegdamsvej DK-2100 Copenhagen, Denmark