EDITORIAL
Chemotherapy for Advanced Aggressive Lymphoma: More Is Better ...
IT
IS AMAZING how difficult it can be to prove something that seems intuitively obvious. Chemotherapeutic agents are directly toxic to tumor cells, and the degree of killing is usually directly proportional to the dose of the agent and the duration of exposure. Thus, it is intuitively obvious that more drug is better. But wait. Most dose-response curves have a plateau at which more drug does not kill more cells. In vivo dose-response curves are affected by tumor burden. Larger tumors require more of an active agent to be eradicated than do smaller tumors. Generally, it is the lack of specificity of the drug that produces toxicity to normal tissues, and it is the toxicity to normal tissues that limits our capacity to further escalate doses. Efforts to increase doses in the face of toxicity can yield the ultimate Pyrrhic victory-death free of cancer but due to complications of treatment. The therapeutic index is often quite narrow for chemotherapeutic agents. The important clinical question is whether we have reached a plateau of the tumor dose-response curve when we administer the so-called maximum-tolerated dose. If we could ameliorate the dose-limiting normal tissue toxicity, would further dose escalations improve treatment outcome? A growing body of data suggests that for some tumors they would.' Aggressive lymphoma may be such a tumor. Evidence that dose intensity is an important determinant of treatment outcome in aggressive histology lymphoma comes from several sources. However, most of the evidence is circumstantial. There are no prospective randomized studies stratified for known clinical prognostic factors in which dose intensity is the controlled variable. Thus, what we are left with is less than ideal. First, DeVita et al2 devised a method to relate the projected nine-drug relative dose intensities of different treatment regimens to the probability of survival and found a significant correlation between dose intensity and treatment outcome. There are also two prospective randomized studies in which the projected dose intensities of two regimens were different and the more doseintensive regimen produced the better response rate. 3' 4 However, the more dose-intensive regimen in each study contained doxorubicin, while 952
Isn't It?
the less dose-intensive regimen did not. Thus, the difference could be related to the inclusion of a more active drug rather than the dose intensity. One small study involved comparing the outcome of two methods of administering cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP)-bleomycin chemotherapy in nonrandomized selected groups of patients, one of which received conventional-dose CHOP-bleomycin and the other CHOP-bleomycin with a dose-escalation scheme for cyclophosphamide and doxorubicin. 5 The dose intensities of the two regimens were about 20% different, and there was no significant difference in outcome. A German multicenter study compared cyclophosphamide, vincristine, prednisone, bleomycin, doxorubicin, and procarbazine (COP-BLAM) to CHOP in a nonrandomized fashion and found that the more dose-intensive COP-BLAM induced a significantly higher complete response rate and survival.6 Studies using the projected dose intensity as a gauge of efficacy probably have the least sensitivity since the projected dose intensity is rarely fully delivered. Furthermore, such calculations generally consider all agents equally effective and cannot account for possible effects related to drug scheduling. However, Meyer et a17 have performed a metaanalysis based on 14 randomized trials involving 2,366 patients. The probability of achieving complete remission was 1.34 times higher among the pooled patients receiving the higher projected dose intensity. The variable lengths of follow-up made it difficult to compare long-term disease-free survival, the most desirable end point for comparison. A second group of studies that may suggest a role for dose intensity in treatment outcome in aggressive lymphoma are single-arm studies compared with historical controls. The cyclophosphamide, vincristine, methotrexate, leucovorin, and cytarabine (COMLA) regimen involves the administration of cyclophosphamide every 12 weeks, vincristine on days 1 and 8, and weekly cytarabine and methotrexate. 8 The complete response rate from this program was 44%, and about 30% of patients were long-term disease-free survivors. Baer et a19 doubled the dose intensity of the cyclophosphamide in COMLA by administering
Journalof Clinical Oncology, Vol 8, No 6 (June), 1990: pp 952-955
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
953
EDITORIAL
it every 6 weeks instead of every 12 weeks and obtained complete responses in 80% of a small series of patients. Gulati et al' 0 identified a group of 14 patients with poor prognosis (B symptoms, high lactic dehydrogenase[LDH], and large mass) based upon their historical treatment experience with conventional-dose therapy at Memorial Sloan-Kettering Cancer Center and treated them with high-dose cyclophosphamide plus total body irradiation as a component of their primary therapy. The 79% disease-free survival rate was nearly four times higher than that of their historical control group. Of course, there is considerable evidence that about 40% of patients with relapsed aggressive histology lymphoma may be cured with high-dose chemotherapy with or without radiation therapy plus autologous bone marrow and/or peripheral blood-stem cell support (with or without colony-stimulating factors)." The salvage rate of such patients with conventional-dose chemotherapy is less than 10%. In the absence of a prospective controlled trial, the strongest type of data that could imply a role for dose intensity in treatment outcome is generated by the retrospective multifactorial analysis of prognostic factors that include the delivered dose intensity. Three such studies have been reported. Epelbaum et al' 2 found a statistically significant association between the delivered dose intensity of cyclophosphamide and survival in their CHOP-treated patients. Shipp et a1' 3 initially found that patients treated with high- or low-dose methotrexate with leucovorin rescue, bleomycin, doxorubicin, cyclophosphamide, vincristine, and dexamethasone (M- or m-BACOD) who received over 80% of the projected dose of cyclophosphamide, doxorubicin, and vincristine had significantly higher complete response rates and survival. Subsequent analysis of a larger number of patients failed to confirm the effect of dose intensity on treatment outcome.14 Coiffier et al15 found that their patients entered onto the LNH-84 protocol who received therapy with a decreased dose intensity (which was primarily due to treatment delays rather than compromising dose) had a greater relapse rate, but higher dose intensities were associated with a higher rate of toxic deaths. In this issue of the Journalof Clinical Oncology, Kwak et al' 6 perform a retrospective analysis of their results using three different treatment
regimens (CHOP, M-BACOD, methotrexate with leucovorin, doxorubicin, cyclophosphamide, vincristine, prednisone, and bleomycin [MACOPB]) in 115 patients with aggressive histology lymphoma. Using a "classification" technique called multivariate (tree-structured) recursive partitioning to identify prognostic groups, they found that the actual or delivered relative dose intensity of doxorubicin greater than 75% in the first 12 weeks of therapy was the single most important predictor of survival. This is an important study since it describes a method of analyzing data retrospectively that has not previously been applied to the dose-intensity question. Basically a computer examines a number of variables and, in an ordered way, chooses an individual parameter that can be dichotomized to give the best split between patients based on a survival end point. The dichotomized variable that best splits the cohort into a "better outcome" versus a "poorer outcome" is considered the most important factor, and after the first split, other factors are examined in each of the two groups to make further splits. Splits continue to be made on the basis of individual characteristics until there are no more factors that split the prognosis of the remaining groups of patients. Each sequential partitioning is conditioned on the previous categorization. Thus, some patients who originally fall into a better prognosis category may subsequently be demonstrated to fall into a poor prognostic subgroup, and vice versa. There are some small quibbles one can make with the analysis. For example, although some variables (eg, sex, B symptoms) are already dichotomized, other variables are continuous (eg, LDH levels, dose intensity). It would have been useful to know the robustness of the splits of continuous variables. Second, dose intensity is not a pretreatment variable, it is a treatment-related variable. As such, there may well be pretreatment variables (eg, hemoglobin levels, bone marrow involvement, performance status, age) that affect dose intensity. Thus, it is possible that dose intensity is a common surrogate for a number of other variables that do not alone make a significant partitioning impact in their analysis, but together would. Finally, it would have been useful for the authors to have used their multivariate analysis (Table 4 in their report) to generate the proportional hazard function that would
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
954
DAN L. LONGO
allow prediction of the outcome of the next patient or group of patients. Descriptive or classification statistical techniques such as the one used are said to have certain weaknesses in permitting projections since they are influenced greatly by the degree to which the described group is like the subsequent group. Logistic regression techniques mathematically take potential inhomogeneities into account. Yet having made this point, I hasten to add that I feel quite confident that their findings based on this interesting tree-structured partitioning method are generally applicable. I hope the software to perform this type of analysis becomes readily available. In addition to the interesting method of data analysis, the most intriguing hypothesis to emerge from this study is that delivering a higher dose intensity of drugs, particularly doxorubicin, might improve the treatment outcome. However, we need to understand that we may be deluding ourselves to make that intellectual leap. The major reason to make that inference is that it is the only partitioning factor that we control (or think we do). However, because there may be a constellation of pretreatment clinical prognostic factors that contribute to our inability to safely deliver more drug, we must be cautious about guessing that the course of disease can be altered by giving more drug to the poor-prognosis subgroup. It is important to note that the group that fared the poorest in this study (Fig 4 in the report by Kwak et a1'6 ) was comprised of patients with high LDH who had received less than 75% doxorubicin dose intensity plus the group with poor performance status who had received greater than 75% doxorubicin. The dose intensity did not seem to make much of an impact on the latter subset and it may be that increasing dose intensity will not help the next group of patients with these characteristics. What is sorely needed is a prospective randomized trial stratified for known prognostic factors in which the only variable is the dose intensity of the treatment. We have recently completed a study of etoposide, cyclophosphamide, doxorubicin, methotrexate with leucovorin, prednisone, mechlorethamine, vincristine and procarbazine (ProMACE-MOPP) versus mechlorethamine, procarbazine deleted from ProMACE-MOPP and cytarabine and bleomycin added (ProMACECytaBOM) that demonstrated a significant
survival advantage to patients treated with ProMACE-CytaBOM. ProMACE-CytaBOM delivers a higher dose intensity of cyclophosphamide, doxorubicin, etoposide, and vincristine, four drugs shared between the two regimens. However, it is also possible (though, in my view, less likely) that the difference in outcome between the two programs is because bleomycin and cytarabine (in ProMACE-CytaBOM) are more active than mechlorethamine and procarbazine (in ProMACE-MOPP). We have now developed a new regimen called short-course ProMACECytaBOM that delivers cyclophosphamide, doxorubicin, and etoposide in alternate weeks with the CytaBOM agents, analogous to the MACOPB program. In a pilot study, the dose intensity of all the ProMACE-CytaBOM agents actually delivered to the first 24 patients receiving shortcourse ProMACE-CytaBOM is significantly higher than the dose intensity of the agents delivered in our previous study of ProMACECytaBOM. We are planning a prospective randomized trial between ProMACE-CytaBOM and short-course ProMACE-CytaBOM that we hope will address the dose-intensity question in a controlled fashion. Alternatively, it may be necessary, as suggested by Kwak et al,' 6 to identify particularly poor subsets of patients for very high doseintensity treatment with bone marrow support. Such an approach has already been piloted with apparent success.1' Such treatment approaches will require physicians to be prepared to manage serious toxicities. It is important to keep in mind that the so-called maximum-tolerated dose depends on the eye of the beholder. The maximum-tolerated dose is really variable and dependent on an assessment of the risk/benefit ratio in a particular setting. Most of us would set the maximumtolerated dose of an agent lower in a patient with lung or pancreatic cancer than in a patient with acute leukemia or lymphoma. If we believe a drug or a regimen has no potential for cure, we are much less likely to ask a patient to put up with grade 3 toxicity. However, we routinely put patients through life-threatening toxicities if there is a reasonable likelihood for cure and if the toxicities are reversible. Primum non nocere is an anachronistic apothegm for modern medicine in general and oncology in particular. Today we do
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
955
EDITORIAL
not just offer nostrums and passively chronicle the natural history of disease; we intervene, we invade, we alter natural history. A physician cannot cure cancer without doing harm, and the degree to which we are willing to cause harm is directly related to our estimate of a patient's curability. A more appropriate aphorism for modern medicine is primum succurrere: first, hasten to help. If more is better and if more
severe acute, reversible toxicity is the price to be paid for long-term disease-free survival, I believe most patients would be willing to pay it.
Dan L. Longo National Cancer Institute FrederickCancerResearch Facility Frederick,MD
REFERENCES 1. Hryniuk WM: The importance of dose intensity in the outcome of chemotherapy, in DeVita VT Jr, Hellman S, Rosenberg SA (eds): Advances in Oncology. Philadelphia, PA, Lippincott, 1988, pp 121-141 2. DeVita VT Jr, Hubbard SM, Longo DL: The chemotherapy of lymphomas: Looking back, moving forward-The Richard and Hinda Rosenthal Foundation award lecture. Cancer Res 47:5810-5824, 1987 3. Hagberg H, Bjorkholm M, Glielius B, et al: CHOP vs MEV for the treatment of non-Hodgkin's lymphoma of unfavourable histopathology: A randomized trial. Eur J Cancer Clin Oncol 21:175-179, 1985 4. Gams RA, Rainey M, Dandy M, et al: Phase III study of BCOP vs CHOP in unfavorable categories of malignant lymphoma: A Southeastern Cancer Study Group trial. J Clin Oncol 3:1188-1195, 1985 5. Lee R, Cabanillas F, Bodey GP, et al: A 10-year update of CHOP-bleo in the treatment of diffuse large-cell lymphoma. J Clin Oncol 4:1455-1461, 1986 6. Gerhartz HH, Thiel E, Hiller E, et al: CHOP and COPBLAM chemotherapy for diffuse large cell nonHodgkin's lymphomas: A retrospective comparison. Hematol Oncol 6:13-19, 1988 7. Meyer RM, Hryniuk WM, Goodyear MDE: The role of dose intensity in determining outcome in intermediate grade non-Hodgkin's lymphoma. Ann Intern Med (submitted) 8. Gaynor ER, Ultmann JE, Golomb HM, et al: Treatment of diffuse histiocytic lymphoma (DHL) with COMLA (cyclophosphamide, oncovin, methotrexate, leucovorin, cytosine arabinoside): A 10-year experience in a single institution. J Clin Oncol 3:1596-1604, 1985
9. Baer MR, Stein RS, Greer JP, et al: Modified cyclophosphamide, vincristine, methotrexate, leucovorin, cytarabine (COMLA) in intermediate- and high-grade lymphoma: An effective short-course regimen. Cancer Treat Rep 70:785787, 1986 10. Gulati SC, Shank B, Black P, et al: Autologous bone marrow transplantation for patients with poor prognosis lymphoma. J Clin Oncol 6:1303-1313, 1988 11. Armitage JO: Bone marrow transplantation in patients with lymphoma. Blood 73:1749-1758, 1989 12. Epelbaum R, Haim N, Ben-Shahar M, et al: Dose intensity for CHOP chemotherapy in diffuse aggressive large cell lymphoma. Isr J Med Sci 24:533-537, 1988 13. Shipp M, Klatt M, Harrington D, et al: M-BACOD and m-BACOD in the treatment of unfavorable prognosis lymphoma: Analysis of prognostic variables. Proc Am Soc Clin Oncol 4:205, 1985 (abstr C-799) 14. Shipp MA, Harrington DP, Klatt MM, et al: Identification of major prognostic subgroups of patients with largecell lymphoma treated with m-BACOD or M-BACOD. Ann Intern Med 104:757-765, 1986 15. Coiffier B, Gisselbrecht C, Herbrecht R, et al: LNH84 regimen: A multicenter study of intensive chemotherapy in 737 patients with aggressive malignant lymphoma. J Clin Oncol 7:1018-1026, 1989 16. Kwak LW, Halpern J, Olshen RA, et al: Prognostic significance of actual dose intensity in diffuse large-cell lymphoma: Results of a tree-structured survival analysis. J Clin Oncol 8:963-977, 1990
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
Chemotherapy for Advanced Aggressive Lymphoma: More Is Better ...
IT
IS AMAZING how difficult it can be to prove something that seems intuitively obvious. Chemotherapeutic agents are directly toxic to tumor cells, and the degree of killing is usually directly proportional to the dose of the agent and the duration of exposure. Thus, it is intuitively obvious that more drug is better. But wait. Most dose-response curves have a plateau at which more drug does not kill more cells. In vivo dose-response curves are affected by tumor burden. Larger tumors require more of an active agent to be eradicated than do smaller tumors. Generally, it is the lack of specificity of the drug that produces toxicity to normal tissues, and it is the toxicity to normal tissues that limits our capacity to further escalate doses. Efforts to increase doses in the face of toxicity can yield the ultimate Pyrrhic victory-death free of cancer but due to complications of treatment. The therapeutic index is often quite narrow for chemotherapeutic agents. The important clinical question is whether we have reached a plateau of the tumor dose-response curve when we administer the so-called maximum-tolerated dose. If we could ameliorate the dose-limiting normal tissue toxicity, would further dose escalations improve treatment outcome? A growing body of data suggests that for some tumors they would.' Aggressive lymphoma may be such a tumor. Evidence that dose intensity is an important determinant of treatment outcome in aggressive histology lymphoma comes from several sources. However, most of the evidence is circumstantial. There are no prospective randomized studies stratified for known clinical prognostic factors in which dose intensity is the controlled variable. Thus, what we are left with is less than ideal. First, DeVita et al2 devised a method to relate the projected nine-drug relative dose intensities of different treatment regimens to the probability of survival and found a significant correlation between dose intensity and treatment outcome. There are also two prospective randomized studies in which the projected dose intensities of two regimens were different and the more doseintensive regimen produced the better response rate. 3' 4 However, the more dose-intensive regimen in each study contained doxorubicin, while 952
Isn't It?
the less dose-intensive regimen did not. Thus, the difference could be related to the inclusion of a more active drug rather than the dose intensity. One small study involved comparing the outcome of two methods of administering cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP)-bleomycin chemotherapy in nonrandomized selected groups of patients, one of which received conventional-dose CHOP-bleomycin and the other CHOP-bleomycin with a dose-escalation scheme for cyclophosphamide and doxorubicin. 5 The dose intensities of the two regimens were about 20% different, and there was no significant difference in outcome. A German multicenter study compared cyclophosphamide, vincristine, prednisone, bleomycin, doxorubicin, and procarbazine (COP-BLAM) to CHOP in a nonrandomized fashion and found that the more dose-intensive COP-BLAM induced a significantly higher complete response rate and survival.6 Studies using the projected dose intensity as a gauge of efficacy probably have the least sensitivity since the projected dose intensity is rarely fully delivered. Furthermore, such calculations generally consider all agents equally effective and cannot account for possible effects related to drug scheduling. However, Meyer et a17 have performed a metaanalysis based on 14 randomized trials involving 2,366 patients. The probability of achieving complete remission was 1.34 times higher among the pooled patients receiving the higher projected dose intensity. The variable lengths of follow-up made it difficult to compare long-term disease-free survival, the most desirable end point for comparison. A second group of studies that may suggest a role for dose intensity in treatment outcome in aggressive lymphoma are single-arm studies compared with historical controls. The cyclophosphamide, vincristine, methotrexate, leucovorin, and cytarabine (COMLA) regimen involves the administration of cyclophosphamide every 12 weeks, vincristine on days 1 and 8, and weekly cytarabine and methotrexate. 8 The complete response rate from this program was 44%, and about 30% of patients were long-term disease-free survivors. Baer et a19 doubled the dose intensity of the cyclophosphamide in COMLA by administering
Journalof Clinical Oncology, Vol 8, No 6 (June), 1990: pp 952-955
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
953
EDITORIAL
it every 6 weeks instead of every 12 weeks and obtained complete responses in 80% of a small series of patients. Gulati et al' 0 identified a group of 14 patients with poor prognosis (B symptoms, high lactic dehydrogenase[LDH], and large mass) based upon their historical treatment experience with conventional-dose therapy at Memorial Sloan-Kettering Cancer Center and treated them with high-dose cyclophosphamide plus total body irradiation as a component of their primary therapy. The 79% disease-free survival rate was nearly four times higher than that of their historical control group. Of course, there is considerable evidence that about 40% of patients with relapsed aggressive histology lymphoma may be cured with high-dose chemotherapy with or without radiation therapy plus autologous bone marrow and/or peripheral blood-stem cell support (with or without colony-stimulating factors)." The salvage rate of such patients with conventional-dose chemotherapy is less than 10%. In the absence of a prospective controlled trial, the strongest type of data that could imply a role for dose intensity in treatment outcome is generated by the retrospective multifactorial analysis of prognostic factors that include the delivered dose intensity. Three such studies have been reported. Epelbaum et al' 2 found a statistically significant association between the delivered dose intensity of cyclophosphamide and survival in their CHOP-treated patients. Shipp et a1' 3 initially found that patients treated with high- or low-dose methotrexate with leucovorin rescue, bleomycin, doxorubicin, cyclophosphamide, vincristine, and dexamethasone (M- or m-BACOD) who received over 80% of the projected dose of cyclophosphamide, doxorubicin, and vincristine had significantly higher complete response rates and survival. Subsequent analysis of a larger number of patients failed to confirm the effect of dose intensity on treatment outcome.14 Coiffier et al15 found that their patients entered onto the LNH-84 protocol who received therapy with a decreased dose intensity (which was primarily due to treatment delays rather than compromising dose) had a greater relapse rate, but higher dose intensities were associated with a higher rate of toxic deaths. In this issue of the Journalof Clinical Oncology, Kwak et al' 6 perform a retrospective analysis of their results using three different treatment
regimens (CHOP, M-BACOD, methotrexate with leucovorin, doxorubicin, cyclophosphamide, vincristine, prednisone, and bleomycin [MACOPB]) in 115 patients with aggressive histology lymphoma. Using a "classification" technique called multivariate (tree-structured) recursive partitioning to identify prognostic groups, they found that the actual or delivered relative dose intensity of doxorubicin greater than 75% in the first 12 weeks of therapy was the single most important predictor of survival. This is an important study since it describes a method of analyzing data retrospectively that has not previously been applied to the dose-intensity question. Basically a computer examines a number of variables and, in an ordered way, chooses an individual parameter that can be dichotomized to give the best split between patients based on a survival end point. The dichotomized variable that best splits the cohort into a "better outcome" versus a "poorer outcome" is considered the most important factor, and after the first split, other factors are examined in each of the two groups to make further splits. Splits continue to be made on the basis of individual characteristics until there are no more factors that split the prognosis of the remaining groups of patients. Each sequential partitioning is conditioned on the previous categorization. Thus, some patients who originally fall into a better prognosis category may subsequently be demonstrated to fall into a poor prognostic subgroup, and vice versa. There are some small quibbles one can make with the analysis. For example, although some variables (eg, sex, B symptoms) are already dichotomized, other variables are continuous (eg, LDH levels, dose intensity). It would have been useful to know the robustness of the splits of continuous variables. Second, dose intensity is not a pretreatment variable, it is a treatment-related variable. As such, there may well be pretreatment variables (eg, hemoglobin levels, bone marrow involvement, performance status, age) that affect dose intensity. Thus, it is possible that dose intensity is a common surrogate for a number of other variables that do not alone make a significant partitioning impact in their analysis, but together would. Finally, it would have been useful for the authors to have used their multivariate analysis (Table 4 in their report) to generate the proportional hazard function that would
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
954
DAN L. LONGO
allow prediction of the outcome of the next patient or group of patients. Descriptive or classification statistical techniques such as the one used are said to have certain weaknesses in permitting projections since they are influenced greatly by the degree to which the described group is like the subsequent group. Logistic regression techniques mathematically take potential inhomogeneities into account. Yet having made this point, I hasten to add that I feel quite confident that their findings based on this interesting tree-structured partitioning method are generally applicable. I hope the software to perform this type of analysis becomes readily available. In addition to the interesting method of data analysis, the most intriguing hypothesis to emerge from this study is that delivering a higher dose intensity of drugs, particularly doxorubicin, might improve the treatment outcome. However, we need to understand that we may be deluding ourselves to make that intellectual leap. The major reason to make that inference is that it is the only partitioning factor that we control (or think we do). However, because there may be a constellation of pretreatment clinical prognostic factors that contribute to our inability to safely deliver more drug, we must be cautious about guessing that the course of disease can be altered by giving more drug to the poor-prognosis subgroup. It is important to note that the group that fared the poorest in this study (Fig 4 in the report by Kwak et a1'6 ) was comprised of patients with high LDH who had received less than 75% doxorubicin dose intensity plus the group with poor performance status who had received greater than 75% doxorubicin. The dose intensity did not seem to make much of an impact on the latter subset and it may be that increasing dose intensity will not help the next group of patients with these characteristics. What is sorely needed is a prospective randomized trial stratified for known prognostic factors in which the only variable is the dose intensity of the treatment. We have recently completed a study of etoposide, cyclophosphamide, doxorubicin, methotrexate with leucovorin, prednisone, mechlorethamine, vincristine and procarbazine (ProMACE-MOPP) versus mechlorethamine, procarbazine deleted from ProMACE-MOPP and cytarabine and bleomycin added (ProMACECytaBOM) that demonstrated a significant
survival advantage to patients treated with ProMACE-CytaBOM. ProMACE-CytaBOM delivers a higher dose intensity of cyclophosphamide, doxorubicin, etoposide, and vincristine, four drugs shared between the two regimens. However, it is also possible (though, in my view, less likely) that the difference in outcome between the two programs is because bleomycin and cytarabine (in ProMACE-CytaBOM) are more active than mechlorethamine and procarbazine (in ProMACE-MOPP). We have now developed a new regimen called short-course ProMACECytaBOM that delivers cyclophosphamide, doxorubicin, and etoposide in alternate weeks with the CytaBOM agents, analogous to the MACOPB program. In a pilot study, the dose intensity of all the ProMACE-CytaBOM agents actually delivered to the first 24 patients receiving shortcourse ProMACE-CytaBOM is significantly higher than the dose intensity of the agents delivered in our previous study of ProMACECytaBOM. We are planning a prospective randomized trial between ProMACE-CytaBOM and short-course ProMACE-CytaBOM that we hope will address the dose-intensity question in a controlled fashion. Alternatively, it may be necessary, as suggested by Kwak et al,' 6 to identify particularly poor subsets of patients for very high doseintensity treatment with bone marrow support. Such an approach has already been piloted with apparent success.1' Such treatment approaches will require physicians to be prepared to manage serious toxicities. It is important to keep in mind that the so-called maximum-tolerated dose depends on the eye of the beholder. The maximum-tolerated dose is really variable and dependent on an assessment of the risk/benefit ratio in a particular setting. Most of us would set the maximumtolerated dose of an agent lower in a patient with lung or pancreatic cancer than in a patient with acute leukemia or lymphoma. If we believe a drug or a regimen has no potential for cure, we are much less likely to ask a patient to put up with grade 3 toxicity. However, we routinely put patients through life-threatening toxicities if there is a reasonable likelihood for cure and if the toxicities are reversible. Primum non nocere is an anachronistic apothegm for modern medicine in general and oncology in particular. Today we do
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
955
EDITORIAL
not just offer nostrums and passively chronicle the natural history of disease; we intervene, we invade, we alter natural history. A physician cannot cure cancer without doing harm, and the degree to which we are willing to cause harm is directly related to our estimate of a patient's curability. A more appropriate aphorism for modern medicine is primum succurrere: first, hasten to help. If more is better and if more
severe acute, reversible toxicity is the price to be paid for long-term disease-free survival, I believe most patients would be willing to pay it.
Dan L. Longo National Cancer Institute FrederickCancerResearch Facility Frederick,MD
REFERENCES 1. Hryniuk WM: The importance of dose intensity in the outcome of chemotherapy, in DeVita VT Jr, Hellman S, Rosenberg SA (eds): Advances in Oncology. Philadelphia, PA, Lippincott, 1988, pp 121-141 2. DeVita VT Jr, Hubbard SM, Longo DL: The chemotherapy of lymphomas: Looking back, moving forward-The Richard and Hinda Rosenthal Foundation award lecture. Cancer Res 47:5810-5824, 1987 3. Hagberg H, Bjorkholm M, Glielius B, et al: CHOP vs MEV for the treatment of non-Hodgkin's lymphoma of unfavourable histopathology: A randomized trial. Eur J Cancer Clin Oncol 21:175-179, 1985 4. Gams RA, Rainey M, Dandy M, et al: Phase III study of BCOP vs CHOP in unfavorable categories of malignant lymphoma: A Southeastern Cancer Study Group trial. J Clin Oncol 3:1188-1195, 1985 5. Lee R, Cabanillas F, Bodey GP, et al: A 10-year update of CHOP-bleo in the treatment of diffuse large-cell lymphoma. J Clin Oncol 4:1455-1461, 1986 6. Gerhartz HH, Thiel E, Hiller E, et al: CHOP and COPBLAM chemotherapy for diffuse large cell nonHodgkin's lymphomas: A retrospective comparison. Hematol Oncol 6:13-19, 1988 7. Meyer RM, Hryniuk WM, Goodyear MDE: The role of dose intensity in determining outcome in intermediate grade non-Hodgkin's lymphoma. Ann Intern Med (submitted) 8. Gaynor ER, Ultmann JE, Golomb HM, et al: Treatment of diffuse histiocytic lymphoma (DHL) with COMLA (cyclophosphamide, oncovin, methotrexate, leucovorin, cytosine arabinoside): A 10-year experience in a single institution. J Clin Oncol 3:1596-1604, 1985
9. Baer MR, Stein RS, Greer JP, et al: Modified cyclophosphamide, vincristine, methotrexate, leucovorin, cytarabine (COMLA) in intermediate- and high-grade lymphoma: An effective short-course regimen. Cancer Treat Rep 70:785787, 1986 10. Gulati SC, Shank B, Black P, et al: Autologous bone marrow transplantation for patients with poor prognosis lymphoma. J Clin Oncol 6:1303-1313, 1988 11. Armitage JO: Bone marrow transplantation in patients with lymphoma. Blood 73:1749-1758, 1989 12. Epelbaum R, Haim N, Ben-Shahar M, et al: Dose intensity for CHOP chemotherapy in diffuse aggressive large cell lymphoma. Isr J Med Sci 24:533-537, 1988 13. Shipp M, Klatt M, Harrington D, et al: M-BACOD and m-BACOD in the treatment of unfavorable prognosis lymphoma: Analysis of prognostic variables. Proc Am Soc Clin Oncol 4:205, 1985 (abstr C-799) 14. Shipp MA, Harrington DP, Klatt MM, et al: Identification of major prognostic subgroups of patients with largecell lymphoma treated with m-BACOD or M-BACOD. Ann Intern Med 104:757-765, 1986 15. Coiffier B, Gisselbrecht C, Herbrecht R, et al: LNH84 regimen: A multicenter study of intensive chemotherapy in 737 patients with aggressive malignant lymphoma. J Clin Oncol 7:1018-1026, 1989 16. Kwak LW, Halpern J, Olshen RA, et al: Prognostic significance of actual dose intensity in diffuse large-cell lymphoma: Results of a tree-structured survival analysis. J Clin Oncol 8:963-977, 1990
Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 29, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
