Internet & Web

EVALUATION OF SURVIVAL DATA FOR CHRONIC MYELOCYTIC LEUKEMIA

Categories
Published
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Description
American Journal of Hematology 1: (1976) EVALUATION OF SURVIVAL DATA FOR CHRONIC MYELOCYTIC LEUKEMIA Joseph E. Sokal, M.D. Department of Medicine B, Rosweil Park Memorial Institute, New York State
Transcript
American Journal of Hematology 1: (1976) EVALUATION OF SURVIVAL DATA FOR CHRONIC MYELOCYTIC LEUKEMIA Joseph E. Sokal, M.D. Department of Medicine B, Rosweil Park Memorial Institute, New York State Department of Health, Buffalo Median survival figures are essentially worthless for evaluation of the effectiveness of therapy in different series of patients with chronic myelocytic leukemia. Such data are heavily influenced by the characteristics of the initial patient mix, which play the major role in determining mortality during the first year of treatment. The effects of differing initial proportions of poor-risk patients are dissipated within two years, however, and the annual mortality rates thereafter reflect the influence of antileukemic therapy. Comparison of these rates, from very diverse series of patients, yielded quite consistent results and indicated superiority of effective chemotherapy over irradiation or inferior drug schedules. Key words: chronic myelocytic leukemia, survival INTRODUCTION Chronic myelocytic leukemia (CML) is one of the more lethal malignant neoplasms, with no cures reported and very few long-term survivors. Despite the fact that good to excellent remissions can be achieved in a majority of patients with any of several therapeutic agents introduced during the past 25 years, relatively little improvement in survival has been recorded during this period. The course of the disease appears to be determined principally by intrinsic factors rather than by antileukemic treatment, and the therapeutic agents now available provide only a little delay in its fatal evolution (1,2). Evaluation of the effect of treatment on the course of CML is complicated by the rather wide variation in survival data reported from different sources. Median survival of untreated patients was about 2.5 years from onset of symptoms (3), while that of treated patients ranges from less than 1 year (4) to about 3.5 years (5,6) from diagnosis. The differences among different series are clearly greater than any possible survival advantage attributable to antileukemic therapy. Received June 18,1976, accepted July 23, Address reprint requests to Joseph E. Sokal, M.D., Medicine B, Roswell Park Memorial Institute, 666 Elm Street, Buffalo, NY Alan R. Liss, Inc., 150 Fifth Avenue, New York, NY 10011 494 Concise Reviews: Sokal loo I ao 194 PHI-POS. PTS W I Q I- z 30 w u a W a I I YEARS FROM DIAGNOSIS Fig. 1. Survival curves for: NCI) a selected population of referred patients, all Ph -positive; RPMI) an unselected population of referred patients; End Results Group) an unselected population of primary patients. All patients received chemotherapy, but agents and schedules differed. Death rates were stable, and similar in all three series, after 2 years. SOME SURVIVAL CURVES IN CML Figure 1 presents three survival plots for patients receiving chemotherapy as their principal treatment modality. At one extreme is a series of 194 ph -positive patients studied at the National Cancer Institute (7). This is a selected series; the referral and acceptance process, and karyotypic classification, excluded a variety of high-risk patients. Death rate during the first year after diagnosis was only 8%. Mortality increased during the second year and then became stable, at 26% per year. At the opposite extreme is a series of 878 patients reported by the End Results Group (8). This series represents a primary patient rather than a referral patient population, with no exclusions. Death rate during the first year after diagnosis was 47%. Mortality decreased during the second year and then became stable, also at 26% per year. The intermediate curve presents the survival of 172 patients studied at our institute. This is a referral population, but reflects no selection on our part; all referrals were accepted. It includes Ph -negative patients and patients who were already in blastic crisis at the time of diagnosis. It is noteworthy that the only significant differences between these survival curves are those recorded during the first two years, and it is evident that these reflect differing initial proportions of poor-risk patients. The data from the End Results Group indicate that during that decade of patient entry ( ), a large percentage of patients with Concise Reviews: Evaluation of Survival Data for CML 49 5 CML were already in the terminal or preterminal stages of the disease at the time that the diagnosis was established. These patients died during the first year. Slight additional excess mortality was recorded during the second year, after which the death rate was stable. In contrast, the selected series from the National Cancer Institute had a low initial death rate, which increased subsequently. Although the Roswell Park series involved no selection on our part, and probably includes a higher proportion of Ph -negative cases than the general population of patients with CML (9), the initial death rate was not as high as that among the patients reported by the End Results Group. This suggests that patients in the terminal stages of leukemia are less likely to be referred to specialized centers than those with a better outlook. Ths conclusion was also reached by Feinleib and MacMahon (4). Presumably, in view of the well-known failure of various therapeutic schedules to improve survival in the terminal phase of CML (10,l l), primary physicians feel that referral to a distant center is of little value in such cases, The stable and virtually identical death rates in the series shown in Fig. 1, after the second year, suggest that a) by that time, all three had been reduced to similar populations of patients and b) the final series of events leading to a fatal termination were then occurring randomly. It should be noted that there were major differences in chemotherapeutic schedules used for these three populations. The great majority of the patients reported by the End Results Group received conventional treatment with busulfan. Most of our patients were treated with other agents, or with investigational two-drug combinations. The National Cancer Institute series includes both conventionally treated patients and investigational routines different from ours. The similar death rates in all three series support the consensus of many authorities that there are no significant differences among various effective chemotherapeutic agents, insofar as survival is concerned. The low initial death rate that increases subsequently, illustrated by the National Cancer Institute series in Fig. 1, is typical of good-risk patient populations (i.e., series containing relatively few end-stage or Ph -negative patients). Almost all data from referral centers and from therapeutic trials with reasonably strict eligibility criteria fit this pattern (see Tables I and 11), while most primary series of unselected patients have relatively high death rates during the first year. An instructive exception to the latter is provided in a report from the United States Air Force (12). Ths deals with a primary patient population, with a one-year death rate of only 15%. Presumably this reflects the earlier diagnosis, which would be expected in a military service setting, with physical fitness requirements and mandatory periodic medical examinations. Thus, it seems reasonable to conclude that the differences between reports from primary and referral centers reflect not differences in the nature of CML seen at different institutions, but differing initial proportions of endstage patients. The natural course of typical CML is characterized by an initial phase of relatively low risk, followed by increasing mortality. This is consistent with the hypothesis that a series of somatic mutations takes place, rather than a single event, before the final, lethal clone of leukemic cells becomes established. Most series using radiation therapy of various types, and the only untreated series in the literature (3), have similar survival curves (Fig. 2). All four of the curves illustrated in this figure fit the pattern for good-risk populations described above. Of interest is the fact that the curves for radiation-treated patients reported by Minot in 1924 and by the British Medical Research Council, more than forty years later, are quite similar. The apparent modest advantage of the former is due to the fact that survival is calculated from onset of symptoms rather than from treatment, as in the latter. 49 6 Concise Reviews: Sokal TABLE I. Survival Data for Patients With CML Receiving Irradiation or No Therapy Mean annual Author or No. Percent alive at: death rate, institution pts. Therapy 1 yr 2yr 3 yr 4 yr 5 yr 2-5 years Osgood (15) 114 Titrated 32P or spray X-ray Principally X-ray X-ray, radium Shimkin et al. (17) Minot et al. (3) Reinhard et al. (1 3) Mayo Clinic (18) Minot et al. (3) Murphy and Howard (19) Med. Research Council (14) P X-ray to spleen None X-ray X-ray to spleen % * * * * % 39% 41 % 41 % 44% 49% 58% *Survival from onset of symptoms attributed to CML. In other series, figures reflect survival from diagnosis or first treatment. TABLE 11. Survival Data for Patients With CML Receiving Chemotherapy Mean annual Author or No. Percent alive at: death rate, institution pts. Therapy 1 yr 2yr 3 yr 4 yr 5 yr 2-5 years Conrad (12) 70 Mayo Clinic (18) 60 End Results Group (8) 868 National Cancer 194 Inst. (7) Ph' + Medical Res. 48 Council (14) Roswell Park 172 Mem. Inst. Dibromomannitol 122 Coop. Study Group (5) Monfardini et al. (20) 94 S chwarzenberg 43 et al. (21) Busulfan Busulfan Principally busulfan Various % 24 % 26% 26% Busulfan % Various % Dibromo mannitol Various, including ineffective* Hydroxyurea, intermittent* 28% 39% 40% *See text for details. In one respect, the curves in Fig. 2 deviate from the pattern illustrated in Fig. 1 ; death rates continue to increase rather than stabilizing after a year or two. Discussion of possible reasons for this difference is beyond the scope of this review. It should be noted, however, that this is not a uniform feature of series utilizing irradiation. Osgood (1 5) treated 114 patients with 32P or total body X-ray irradiation and recorded an essentially stable annual death rate (see Table I). Concise Reviews: Evaluation of Survival Data for CML 49 7 X-RAY OR RADIUM TREATED PTS. 54 PTS.. X-RAY C MED. RES. COUNCIL, SURVIVAL FROM ONSET OF SYMPTOMS I YEARS Fig. 2. Survival curves for: Minot) untreated patients and patients receiving X-ray or radium irradiation (survival from onset of symptoms); Reinhard) patients receiving radioactive phosphate (survival from diagnosis); Medical Research Council) patients treated with splenic irradiation (survival from treatment). Death rates were similar in all four series, but increased progressively through the fourth or fifth year. EVALUATION OF SURVIVAL DATA It is obvious that comparison of median survival figures is essentially worthless for evaluation of the effectiveness of therapy in different series of patients with CML. On the other hand, the slope of the survival curve after the first two years should provide a reliable indication of the efficacy of a particular treatment schedule. Tables I and I1 demonstrate application of this approach. Survival data from a number of the larger reported series are summarized according to type of treatment used. Techniques of calculating and expressing survival were not uniform throughout these reports. In some instances, the actual survival figures were available; in others, these were estimated from the published survival curves, which introduced some inaccuracy. Thus, the data from different series are not comparable in all respects, and some of the values listed in the tables are approximations. Nonetheless, it is evident that consistent patterns can be identified. With the exception of one series (15), radiation was not associated with a significant reduction in death rate over that recorded in the only available untreated series (Table I). Minot (3) commented on the rather surprising fact that although substantial functional and hematologic improvement was produced, irradiation did not prolong survival. Many authors have made similar observations subsequently, and it is principally statistics from series utilizing radiation therapy that have led to the common teaching that antileukemic therapy produces unequivocal hematologic and clinical benefit but does not affect survival. As Bergsagel (16) pointed out some years ago, this statement is inaccurate. 498 Concise Reviews: Sokal Most series utilizing chemotherapy record mean annual death rates below 30% after the second year (Table 11). Both of the exceptions, with mean death rates of 39% and 40%, are rather unusual. One (20) includes patients treated with agents now recognized as clearly inferior (e.g., urethane, chlorambucil, hadacidin, Fowler s solution). In the other (21) hydroxyurea, an agent of recognized effectiveness, was used but patients were allowed to relapse repeatedly to WBC counts as hgh as ( ) X 103/mm3 in order to use them as granulocyte donors. The rather rapid increase in mortality in this series, after the second year, suggests that such a scheme of management is disadvantageous on a longterm basis. It is clear that several schedules of chemotherapy are associated with better survival than is achieved with no treatment, conventional x-ray or 32P irradiation, or chemotherapy with inferior drugs or treatment schedules. The annual mortality rate is reduced by about one-third with such treatment. Busulfan is at least as effective as dibromomannitol or the various other agents used at the National Cancer Institute and at our institution. COMMENTS Considerable confusion has resulted from the fact that different series of patients with CML contain differing initial proportions of high-risk cases. Together with the common practice of using median survival to characterize a series, this has made it difficult to make any sense out of survival reports for this disease. It is evident that when survival data are examined in more detail: a) the effects of more favorable or less favorable initial patient input can be identified, and b) the efficacy of treatment can be examined by a technique that is largely unaffected by the characteristics of the initial pstient mix. The rather close agreement in annual death rates recorded after the second year among initially quite dissimilar series of patients treated with effective schedules of chemotherapy (Table 11) is noteworthy. The course of good-risk groups of CML patients treated with busulfan or comparable agents can be defined with some precision. The death rate during the first year after diagnosis is quite low, 5-10%. Mortality increases during the second year and is relatively stable thereafter, at about 26% per year. The results of effective chemotherapy can be distinguished quite easily from those of inferior treatment schedules (Tables I and 11). All series contain patients with atypical CML, whose prognosis is usually poor. It is not entirely clear whether the differences between different series are due to differing proportions of such cases, or of typical cases in the terminal stages of the disease; the latter seems more likely. In any case, poor-risk patients can usually be identified by objective criteria (22) and thus could be excluded from therapeutic trials if it is desired to study a homogeneous good-risk patient population. Such classification is possible both at the time of diagnosis and later. For example, we were able to identify a good-risk group of patients (evaluated at one year after diagnosis) to serve as historical controls for patients in good status at entry into an immunotherapy program (23). No patient in either group died during the first six months of follow-up, and mortality at one year was 7% among controls and 4% among immunized patients. Similarly, different poor-risk groups of patients can be identified. Thus, it should be possible to evaluate the effects of a new therapeutic agent in CML, even if suitable concurrent controls are not available. If the patients entering treatment are appropriately studied and classified, their expected survival with conventional therapy can be predicted with reasonable accuracy to provide a reference standard for the observed Concise Reviews: Evaluation of Survival Data for CML 499 survival experience. This is true of poor-risk as well as good-risk patients. For example, it was possible to conclude from an uncontrolled Acute Leukemia Group B study (1 1) that none of three schedules of chemotherapy tested offered any advantage in management of the terminal phase of CML. The past few years have seen much interest in exploration of new therapeutic approaches to CML. Various investigators have undertaken exploratory trials of new drug schedules, of aggressive treatment programs attempting to achieve cure in early disease, of prophylactic splenectomy, or of immunotherapy. Many of these studies involve recently diagnosed Ph -positive patients in good clinical and hematologic status. In work with such patients, investigators must keep in mind that a one-year survival of 95% and a two-year survival of 80% are within the expected range. In fact, a death rate exceeding 10% during the first year, or 30% during the first two years, would be cause for suspicion that the new treatment is inferior to conventional therapy. After the two-year point, an annual death rate above 30% would be suspect, while one of 20% or less would suggest superiority over current management. REFERENCES 1. Galton DAG: Chemotherapy of chronic myelocytic leukemia. Seminars in Hematology 6: , Stryckmans PA: Current concepts in chronic myelogenous leukemia. Seminars in Hematology 11 ~ , Minot GR, Buckman TE and Isaacs R: Chronic myelogenous leukemia. Age, incidence, duration and benefit derived from irradiation. JAMA 82: , Feinleib M and MacMahon B: Variation in the duration of survival of patients with the chronic leukemias. Blood 15: , Report of Dibromomannitol Cooperative Study Group: Survival of chronic myeloid leukemia patients treated by dibromomannitol. Europ J Cancer 9: , Weil M, Jacquillat CI, Pereira NM, Gemon MF and Tanzer J: Study of prognostic parameters in chronic myelocytic leukemia. Proc Am A Cancer Res and Am Soc Clin Onc 15:87, Canellos GP: Personal communication. 8. End Results in Cancer: Report No. 4. DHEW Publication No. (NIH) , Ezdinli EZ, Sokal JE, Crosswhite L and Sandberg AA: Philadelphia chromosome-positive and -negative chronic myelocytic leukemia. Ann Int Med 72: , Bernard J and Tanzer J: Chronic myelocytic leukemia. In Cancer Medicine. (Holland JF and Frei E, eds.) Philadelphia: Lea and Febiger, 1973, pp Jayes DM, Ellison RR, Glidewell 0, Holland JF and Silver RT: Chemotherapy for the terminal phase of chronic myelocytic leukemia. Cancer Chemother Rep 58: , Conrad FG: Survival in chronic granulocytic leukemia: splenic irradiation vs. busulfan. Arch Int Med 131: , Reinhard EH, Neely CL and Samples DM: Radioactive phosphorus in the treatment of chronic leukemias over a period of 15 years. Ann Int Med 50: , Medical Research Council s Working Party for Therapeutical Trials in Leukaemia: Chronic granulocytic leukaemia: comparison of radiotherapy and busulphan therapy. Brit Med J 1 : , Osgood EE: Treatment of chronic leukemias. J Nuc Med 5: , Bergsagel DE: The chronic leukemias: A review of disease manifestations and the aims of therapy. Canad Med A J 96: , Shimkin MB, Mettier SR and Bierman HR: Myelocytic leukemia: An analysis of incidence, distribution and fatality, Ann Int Med 35: , Petitt MR and Silverstein MN: Personal communication. 19. Murphy WP and Howard I: Roentgen treatment of chronic leukemia. Am J Roent, Rad Ther and Nuc Med 88: , Monfardini S, Gee T, Jerrold F and Clarkson B: Survival in chronic myelogenous leukemia: Influence of treatment and extent of disease at diagnosis. Cancer 31: , 1973. 500 Concise
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks