Long-Term Follow-Up of the Italian Trial of Interferon- Versus Conventional Chemotherapy in Chronic Myeloid Leukemia

INTERFERON-α (IFN-α) was reported to be active in the treatment of chronic myeloid leukemia (CML) in 1983¹ and to induce cytogenetic remissions in 1986.2 In 1986 we enrolled 322 consecutive CML patients in a prospective randomized study of IFN-α versus conventional chemotherapy. Our reports on cytogenetic response and on survival appeared in 1992³ and in 1994⁴ and provided confirmation to the findings5-7 that IFN-α was able to induce cytogenetic remissions and to the suggestion5 that IFN-α treatment prolonged survival as compared with conventional chemotherapy. Subsequently, the therapeutic role of IFN-α in CML was confirmed again in three prospective multicenter trials.8-11 On the other hand the treatment is chronic, may be required for a very long period of time, and can interfere with other treatment procedures. The effect of IFN-α on long-term survival, the duration and the quality of the response, the late treatment complications, and the compliance with chronic treatment are all important issues that were described so far only in the patients who were originally recruited at a single center at M.D. Anderson Hospital (Houston, TX) because that center was the first to apply IFN-α to the treatment of CML on a large basis.5,12 We report on these issues, based on an unselected multicenter series of patients who were prospectively assigned to treatment with IFN-α or with conventional chemotherapy.

The recruitment for this study was opened July 1986 and was terminated July 1988. Approval was obtained from the Institutional Review Boards. Informed consent was provided according to the Declaration of Helsinki. During that 2-year period, 322 patients were assigned at random, with a ratio of 2:1, to receive either IFN-α (218 cases) or conventional chemotherapy (104 controls). In the IFN-α arm, the drug was human recombinant IFN-α2a (Roferon-A; HoffmannLa Roche), 9 million IU (mIU) daily for a minimum of 14 months and indefinitely in case of any cytogenetic response, until tolerated or until progression to accelerated or blastic phase. Chemotherapy could be used at any time in case of poor hematologic control, and after 14 months in case of no cytogenetic response. In the chemotherapy arm, first-line treatment was hydroxyurea in 94 patients and busulfan in 10 patients.

Accelerated or blastic phase was defined by at least two of the following predetermined criteria: (1) more than 10% blast cells or more than 30% blast cells and promyelocytes in the peripheral blood; (2) more than 15% blast cells or more than 50% blast cells and promyelocytes in the marrow aspirate; (3) splenomegaly (>10 cm below the costal margin) with a white blood cell (WBC) count of less than 25 × 10⁹/L; (4) involvement of the central nervous system, bone, lymph nodes, or other extrahematologic sites; and (5) trisomy Ph, trisomy 8, or isochromosome 17. The cytogenetic response was evaluated after 8 and 14 months and yearly thereafter, based on a minimum of 10 evaluable metaphases, and was defined according to the percentage of Ph negative metaphases as none, minimal (1% to 32% Ph neg), minor (33% to 65% Ph neg), major (66% to 99% Ph neg), and complete (100% Ph neg). The hematologic response was defined as complete if blood counts were normal (hemoglobin > 110 g/L, platelet < 500 × 10⁹/L, WBC < 10 × 10⁹/L), if the differential leukocyte count did not contain immature cells, and if the spleen was not palpable. The formulation of Sokal et al13 was used to categorize the patients by risk.

Survival was calculated according to the method of Kaplan and Meier14 from the date of randomization to the date of death or last contact. Follow-up information was available on all patients. At last contact (June 1997), the follow-up of living patients ranged between 95 and 129 months (median, 112 months). During the chronic phase, 38 patients underwent allogeneic bone marrow transplantation (BMT) (12%) and 6 patients were autografted (2%). Allogeneic BMT was allowed by the protocol, whereas autografts were protocol violations. For that reason, all survival calculations and comparisons were made either including, excluding, or censoring the patients who were submitted to transplant. The results were always identical, and the survival data that are reported hereforth include all randomized patients, with censoring for allogeneic transplant. The time to the progression to accelerated or blastic phase was calculated from randomization, with censoring for allogeneic BMT and for death in chronic phase. Analyses and comparisons were made by the chi-square test and Student’s t-test, by the log-rank method and by logistic regression. The Cox proportional hazards model for covariate analysis of censored data on survival15 was used whenever appropriate. P values were two sided. All the calculations were based on the intention-to-treat principle and always included all the randomized patients. To avoid any bias from using time-dependent variables like hematologic or cytogenetic response, landmark analysis at specific time points was performed whenever appropriate.

The status of the patients who were not transplanted is shown in Table1. In the IFN-α arm 56 of 188 patients (30%) were alive and 16 of these patients (8%) maintained a complete or a major cytogenetic remission. In the chemotherapy arm 17 of 96 patients (18%) were alive, with a minor cytogenetic response in one case.

The number of the patients who discontinued the assigned treatment and went off protocol before progression was 114 of 218 (52%) in the IFN-α arm and 37 of 104 (35%) in the chemotherapy arm. The difference was due to the fact that 39 patients (18%) discontinued IFN-α because of the side effects that are listed in Table2. The classic flu-like syndrome including fatigue, fever, musculoskeletal pain, and headache covered 37% of discontinuations, either early or late. Anorexia, nausea, and diarrhea came second, but only during the first 3 years of treatment. Neurologic side effects like depression, amnesia, lethargia, psychosis, reversible coma in one case, and polyneuropathy, mainly sensorial, developed during the first 5 years of treatment and were the third cause of treatment discontinuation.

In the IFN-α arm 84 patients overall (38%) also received chemotherapy before progression to accelerated or blastic phase, either because IFN-α was discontinued for side effects (39 cases) or because it was believed that IFN-α was unable to control the disease (45 cases). In the chemotherapy arm, no option was provided for crossing the treatment to IFN-α, but IFN-α was actually administered to 8 patients, 2 to 70 months after randomization. The overall survival of these 8 patients was long, ranging from 96 to 129 months (median, 106 months), with 2 of 8 alive at last contact.

The dose of IFN-α that was originally scheduled for this study was 9 mIU daily (corresponding to 63 mIU weekly) for 8 months, to be maintained thereafter at maximum tolerated dose if any cytogenetic response was obtained. During the first year, the mean IFN-α weekly dose was significantly higher in the patients who achieved any degree of cytogenetic response than in the others (Table3). Thereafter, in the majority of the nonresponders the dose was decreased to 9 mIU weekly, as it was prescribed by the protocol. In the responders the dose was also adapted, according to tolerance, and from the 5th year on was about 20 mIU weekly (Table 3).

The option for allogeneic BMT from an HLA-compatible family donor was always open and was actually used during the chronic phase in 33 patients, 2 to 65 months after registration (median, 15 months). In another 5 patients an allogeneic BMT was performed later (42 to 107 months after registration) from a matched unrelated or a partially matched related donor. In summary, 38 of 322 patients of any age (12%) and of 172 patients less than 51 years old (22%) received an allogeneic BMT during the chronic phase. Sixteen of the 38 transplanted patients (42%) died of transplant or transplant-related complications, two are alive with leukemia, and the remaining 20 (53%) are alive and in hematologic remission. Two additional patients, both in the IFN-α arm, were transplanted in blastic phase and died of transplant. Six patients (4 in the IFN-α arm and 2 in the chemotherapy arm) were autografted in chronic phase. Three of them died of leukemia and three are alive with leukemia. Ten patients (6 in the IFN-α arm and 4 in the chemotherapy arm) were autografted in accelerated or blastic phase. Seven died of transplant or leukemia, and three are alive with leukemia.

The median survival of the 218 patients who were assigned to IFN-α and of the 104 patients who were assigned to chemotherapy was 76 months and 52 months, respectively (95% CI, 69 to 86 months and 43 to 66 months, log-rank test, P = .002) (Fig1). The proportion of the patients who were projected to be alive after 10 years was 29% (95% CI, 23% to 36%) in IFN-α arm and 17% (95% CI, 9% to 25%) in chemotherapy arm. Figure 1 also shows the rate of the progression from chronic to accelerated or blastic phase. The median time from diagnosis to progression was 74 months with IFN-α and 46 months for chemotherapy (log-rank test, P = .0005).

Figure 2 shows the survival and the progression from chronic to accelerated or blastic phase of Sokal’s low-risk patients as a function of treatment. Figure3 shows the same for Sokal’s intermediate- and high-risk patients. The difference was significant both in low-risk patients (P = .03 for survival and .02 for progression) and in intermediate- and high-risk patients (P = .006 for survival and .002 for progression).

Fourteen patients died in chronic phase; nine of them had been assigned to IFN-α (4%) and five had been assigned to chemotherapy (5%) (Table 4). The deaths occurred either early (3 to 14 months after diagnosis), and were due to infection, cardiac disease, or gastric bleeding; or late (32 to 99 months after diagnosis), and were due to cardiac or vascular disease, or to other cancers. Fatal cardiovascular events were slightly more frequent in IFN-α patients (5 of 218 or 2.3%) than in chemotherapy patients (1 of 104 or 1%). Cancers were slightly more frequent in chemotherapy patients (4 of 104 or 4%) than in IFN-α patients (2 of 218 or 1%). Among the nine IFN-α patients who died in chronic phase, four were in stable complete or major cytogenetic remission.

Prognostic factors for survival were examined in all the patients and in those who had been randomized to receive IFN-α (Table5). In all the patients, many factors were associated with survival, but by multivariate analysis only treatment arm remained statistically very significant, together with age, spleen, and Sokal’s score. In the patients who had been assigned to IFN-α the survival was significantly influenced by Sokal’s score, hematologic response, and cytogenetic response. Figure4 shows a landmark survival analysis of the patients who were assigned to IFN-α according to hematologic response and to cytogenetic response. Both responses were associated with a substantial and very significant prolongation of survival.

A cytogenetic response (Ph neg metaphases, 33% to 100%) was obtained in 70 of 218 cases (32%). The best cytogenetic response was complete (Ph neg, 100%) in 23 cases (10%), was major (Ph neg, 66% to 99%) in 23 cases (10%), and was minor (Ph neg, 33% to 65%) in the remaining 24 cases (11%). The time to achieve the first response ranged from less than 1 year to 7 years (median, 1 year), but the time to achieve the best response was even longer, with a median of 2 years. That was mainly due to the fact that the majority of the complete cytogenetic responses were detected rather late (4 after 1 year, 6 after 2 years, 5 after 3 years, and 8 after 5 years or more). Though many cytogenetic responses were first recorded after more than 1 year, many responders (48 of 70, 68%) received their first response within the first year of treatment, and other 14 patients (20%) were found to have some Ph neg metaphases within the same period. Only 8 of 70 cases (11%) did not show any Ph neg metaphases after the first year of treatment. One of these patients achieved a complete and stable response later, whereas the other 7 achieved only a major or a minor response, and that response was unstable and transient. Table6 shows the course and the fate of 70 patients who achieved a cytogenetic response. Seven of them (10%) were submitted to allogeneic BMT. Of the remaining 63 patients, 4 died in chronic phase (6%), 34 (54%) are alive and in chronic phase, and the remaining 25 (40%) progressed to accelerated or blastic phase and died of leukemia or related complications. The proportion of the cases with progression was negatively related with the quality of the cytogenetic response (13% for complete response, 39% for major response, and 68% for minor response, P = .005) (Table 6). In the 10 patients who progressed to accelerated or blastic phase after a complete or a major cytogenetic response, two modalities of progression were observed: in 5 cases the loss of the cytogenetic response was rapidly followed by an acute blast crisis, with a lymphoid phenotype in 4 of 5 cases; in the other 5 cases, the time from cytogenetic response loss to progression was much longer, ranging between 17 and 73 months, progression to blastic phase was slow, and the terminal phenotype was myeloid.

The main purpose of this updated report of the Italian study is to provide information on the long-term effects of a policy of chronic IFN-α treatment. The first point was to evaluate the effects of IFN-α on long-term survival, and more clearly to answer the question if the survival benefit over conventional chemotherapy would continue in the long term. This point was not settled in prior reports because the difference between conventional chemotherapy and IFN-α does not concern the cure rate, but the cytogenetic response rate. In this study complete and major cytogenetic responses occurred in 46 of 218 patients (21%) and only 9 patients (4%) were still in complete cytogenetic remission after more than 8 years. Although some of these patients were found to be sometimes also in molecular remission, as defined by nonquantitative reverse transcriptase polymerase chain reaction, and although some cases of molecular remission were described by others,16 in the great majority of the patients leukemia cannot be cleared off by current IFN-α treatment.17,18Therefore, this study cannot provide evidence that IFN-α can cure CML, but shows that the survival benefit of IFN-α over conventional chemotherapy did not die out and was maintained in the long term either in the low-risk patients or in the patients with a higher risk. However, whereas the relative benefit of IFN-α over chemotherapy was independent of the risk, the absolute benefit was not, because nearly all the long survivors were in the low-risk group (Figs 3 and 4). The second point was to evaluate the long-term compliance and the long-term side effects of chronic IFN-α treatment. Late treatment-related toxicity was neither more frequent nor different from early toxicity (Table 2). In particular, no more cases of neurologic toxicity were observed. However, it should be noticed that the number of cases “at risk” after 6 years was only 54 and that the mean dose that was administered after that time was 3 mIU daily. Eight patients (4%) died in chronic phase. In one case death was attributed to treatment-related pancytopenia. In the other 7 cases, a direct responsibility of treatment could not be established, but based on the causes of death (Table 4), special attention should be paid to the chronic administration of IFN-α to patients with established risk factors for cardiac or vascular disease. The development of other cancers was of concern, but the proportion was not different in the two treatment arms (1% with IFN-α and 4% with conventional chemotherapy).

This study was planned in 1985 and was started in 1986 with the aim of evaluating if IFN-α would prolong the survival over conventional chemotherapy. The design was not conceived with the purpose of answering other questions like the prediction, the kinetic, and the duration of the response. However, because these questions are important, the data were examined retrospectively and the results suggested that the cytogenetic response to IFN-α was predicted by Sokal’s risk and by hematologic response, and that the great majority of the cytogenetic responses were achieved within 1 year of treatment, though in many cases the responses improved significantly with time. The results also suggested that when the best cytogenetic response was only minor (Ph neg, 33% to 65%), the response was unstable and of a short duration, whereas the median duration of complete and major responses (Ph neg, ≥66% to 100%) was of about 60 months. These data fit with the observations from the M.D. Anderson Hospital trial.5,12,19 

The treatment protocol prescribed a daily IFN-α dose of 9 mIU, corresponding to 63 mIU weekly to be administered for at least 14 months and indefinitely in the case of any cytogenetic response. The doses that were actually administered ranged widely, with a mean value decreasing from 50 mIU weekly during the first year to about 20 mIU weekly from the 5th year and continuing for the remainder of the study. These data are the result of a treatment policy that was targeted to the maximum tolerated dose and do not answer the questions regarding whether lower doses of IFN-α can be more cost-effective8,9,11,19,20 and if the treatment should be continued forever or could be discontinued once a durable complete or major cytogenetic response has been achieved, as suggested by the observations from M.D. Anderson Hospital.12,19,21 

Evaluating and discussing the long-term effect of IFN-α treatment raises two main problems. The first problem is how to improve the cytogenetic response rate and how to prolong survival even more. The addition of low-dose arabinosyl cytosine was shown to be effective in one study22 and is currently being tested by our group. The second problem concerns the relationship with BMT. Autologous BMT can help to improve survival but there is no evidence of a cure,23,24 and although it is conceivable that the longer the time from diagnosis, the worse the results,25,26pretreating with IFN-α does not prevent from successful autografting.24-27 Pretreating with IFN-α does not prevent from allogeneic BMT either, although delaying the transplant would probably increase transplant-related mortality.28 One study raised the suspicion that IFN-α treatment may specifically adversely affect transplant outcome,29 but more studies showed that this was not the case.30-33 Allogeneic BMT from an HLA-identical sibling is currently considered as the first therapeutic option in the management of CML for patients under the age of 60, and CML is the first in the list of the indications for an allogeneic transplant from a matched unrelated donor.34,35This is conceivable because allogeneic BMT is the only procedure that can cure CML.19,28,34-36 However, CML is no longer a rapidly fatal disease and about 50% of the cases present with low-risk features. In this study, the 10-year survival of the low-risk patients who were assigned to IFN-α is 47% (95% CI, 36% to 59%) for all of the 91 patients and 64% (95% CI, 47% to 80%) for the 36 patients who achieved a cytogenetic response. It should not be overlooked that these are not the best results that can be obtained at a single specialized center, but rather, are the results of a nationwide multicenter study that pioneered the introduction of IFN-α in Italy.

The active members of the group for this study were: N. Testoni, M.D. Zamagni, G. Martinelli: University of Bologna; D. Damiani, M. Michieli: University of Udine; A. Zaccaria, Ospedale Ravenna; G. Specchia, V. Liso: University of Bari; M. Lazzarino, C. Bernasconi: University of Pavia; E. Montefusco, G. Alimena, F. Mandelli: University La Sapienza, Roma; P. Leoni, S. Rupoli, M. Candela: University of Ancona; A. Nosari, L. Gargantini: Ospedale Niguarda Cà-Granda, Milano; I. Majolino, S. Tringali: Ospedale Cervello, Palermo; A.M. Liberati, F. Grignani, A. Tabilio, M. Martelli: University of Perugia; F. Paolino, M. Bertini: Ospedale Molinette, Torino; A. Di Tucci, G. Broccia: Ospedale Oncologico Businco, Cagliari; F. Leoni, S. Ciolli: University of Firenze; L. Luciano, B. Rotoli: University of Napoli; R. Perricone, A. Cajozzo: University of Palermo; A. Montuoro, A. De Laurenzi: Ospedale San Camillo, Roma; F. Palmieri, E. Volpe: Ospedale Avellino; A. D’Emilio, E. di Bona: Ospedale Vicenza; A. Capucci, T. Izzi: Ospedale Brescia; G.L. Scapoli, G.L. Castoldi: University of Ferrara; M. Lombardo, L. Ruberto: University of Chieti; S. Sica, G. Leone: Università Cattolica, Roma; C. Delfini, G. Nicolini: Ospedale Pesaro; F. Papineschi, E. Benedetti: University of Pisa; F. Gualandi, A.M. Marmont: Ospedale San Martino, Genova; D. Dini, G. Torelli: University of Modena; L. Mangoni, V. Rizzoli: University of Parma; M. Girino, E. Ascari: University of Pavia; C.A. Bodenizza, M. Carotenuto: Casa Sollievo della Sofferenza, San Giovanni Rotondo; A. Di Francesco, D. Quaglino: University of L’Aquila; S. Nardelli, F. Ciccone: Ospedale Latina; E. Miraglia, R. De Biasi: Ospedale Nuovo Pellegrini, Napoli; D. Ferrero, A. Pileri: University of Torino; A. Rambaldi, T. Barbui: Ospedale Bergamo; S. Morandi, C. Bergonzi: Ospedale Cremona; C. De Rosa, R. Cimino: Ospedale Cardarelli, Napoli; F. Ronca, F. Nobile: Ospedale Reggio Calabria; M. Cantonetti, S. Amadori: UniversitàTor Vergata, Roma; A. Gallamini, E. Gallo: Ospedale Cuneo; C. Musolino, G. Squadrito: University of Messina; A. Capaldi, M. Aglietta: Ospedale Mauriziano Umberto I, Torino; G. Pinotti, A. Venco: Ospedale Varese; V. Zagonel, A. Pinto: Centro Regionale di Riferimento Oncologico, Aviano; I. Gentilini, P. Coser: Ospedale Bolzano; P. Guglielmo, R. Giustolisi: University of Catania; M. Pizzuti, F. Ricciuti: Ospedale Potenza.