- Email: [email protected]
Advanced diffuse large-cell lymphoma treated with 12-week combination chemotherapy: Natural history of relapse after initial complete response and prognostic variables defining outcome after relapse
Annals of Oncology 8: 1125-1132, 1997. © 1997 Kluwer Academic Publishers. Printed in the Netherlands.
Original article Advanced diffuse large-cell lymphoma treated with 12-week combination chemotherapy: Natural history of relapse after initial complete response and prognostic variables defining outcome after relapse P. I Hoskins,1 N. Le,2 R. D. Gascoyne,3 R. Klasa,1 T. Shenkier,1 S. O'Reilly1 & J. M. Connors1
Summary Purpose: To define both the natural history of and prognostic factors affecting outcome post relapse from a complete response in advanced stage diffuse large-cell lymphoma. Patients and methods: A total of 468 patients aged 17-74 years received the 12-week duration chemotherapy regimens MACOP-B.VACOP-B and ACOP-12 between 1 April 1981 and 31 December 1995 for advanced stage diffuse large, mixed or immunoblastic lymphoma. Of these 402 entered a complete remission, 97 (24%) of whom subsequently relapsed. Initial staging data, follow-up, and relapse information were analyzed to define the natural history of relapse and also subjected to univariate and multivariate correlation with overall (OS) and failure free survival (FFS). Results: Eleven percent of the relapses were low grade. All other relapses were of intermediate grade with 75% occurring
Introduction
within the first two years, the remainder up until the eleventh year. Median and five-year OS from the time of relapse for intermediate grade relapse were 12 months and 20%; for FFS they were eight months and 18% respectively. Adverse independent factors, for both OS and FFS were: less than one year to relapse, decreasing performance status at relapse, and more than three nodal sites at relapse. Conclusions: Low-grade relapse is not uncommon in patients who initially presented with diffuse large cell lymphoma. As the management of low- and intermediate grade disease is so different biopsy proof of the nature of the relapse is of value. The prognostic factors identified need to be taken into consideration when analyzing results from trials of secondary treatment so as to avoid erroneous conclusions about comparative treatment efficacy. Key words: lymphoma, relapse
tumour mass, high LDH level and non low-grade histology [7, 11, 12]. The relevance of this is that if these factors are not stratified for they could lead to survival differences in salvage trials which may falsely be attributed to treatment. In order to further elucidate the natural history of and prognostic factors related to survival after relapse, we conducted this retrospective study of patients with advanced, diffuse large-cell lymphoma who have all been treated with 12-week duration combination chemotherapy.
Up to 40% of patients with advanced diffuse large-cell or immunoblastic lymphoma who respond to first-line multi-agent chemotherapy relapse [1-6]. It is commonly believed that most, if not all, relapses occur within the first 24 months but there are reports of later relapses. In the M.D. Anderson experience 5% of the relapses occurred more than 30 months from diagnosis, with the longest interval being 170 months [7]. The group from the Dana-Farber found a 7.8% late relapse rate, occurring up to seven years [8]. In both of these series lowgrade (i.e., discordant) relapses were also seen. In the SWOG series using CHOP as the primary chemotherPatients and methods apy relapses were seen up to seven years from diagnosis [9]. These later relapses may be biologically different Twelve-week duration chemotherapy (MACOP-B April 1981 to May from early relapses. There is indirect evidence for such a 1986; VACOP-B June 1986 to April 1993 and ACOP12 May 1993 and biologic difference from clinical data. In the Parma continuing) has been our standard treatment for advanced large cell randomized study of salvage chemotherapies the two- lymphoma since 1981 (Figure 1). These three regimens are based on a year survival rate was 19% if the first relapse occurred backbone of doxorubicin, cyclophosphamide, vincristine and predniThey were equivalent with regard to survival outcomes when within 12 months of the initial diagnosis versus 47% if it sone. analyzed within the subgroupings of the International Prognostic Facwere after 12 months [10]. Other prognostic factors have tors Index and so have been considered as one uniform group for the been identified at relapse as predicting for worse sur- purposes of this paper [13]. vival and these include poor performance status, large The eligibility criteria to receive this chemotherapy were: histologic
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
^Division of Medical Oncology, 2Division of Epidemiology, 3Department of Pathology. British Columbia Cancer Agency, Vancouver Centre and the University of British Columbia, Vancouver, British Columbia
1126 MACOP-B (4/81-5/86)
Table 1. Patient characteristics at initial diagnosis. WEEK 1 2 3 4 5 6 7 8 9 10 11 12
Doxorubicm Cyclophojphmude Vmcnfline (ctp-2.0mg) Blcoraycin Mcthotrexate
50meW 35Omg/m' 1 4 mg/m' 10mg/ra' 400mg/mJ
rv IV
X
X
X
X
X
X
X
X
X
X
rv
X
X
X
IV
X
rv
X X
Number Age (years) Male: female
WEEK 1 2 3 4 5 6 7 8 9 10 11 12
rv
IV IV PO IV IV
X
X
X
X
X
X
X
X
X
X X
X
X X X
X
X
X
X X X
X
X
X
X
X
X
ACOP-12 (5/93 - and continuing)
50mg/mJ 35Omg/ra' 1.2mg/m'
IV IV IV
97 21-68 (median 55) 51:46
Diffuse mixed 61(13%) Diffuse large-cleaved 52(11%) Diffuse large non-cleaved 114(24%) Diffuse large NOS 134(29%) Immunoblastic 107 (23%)
53(13%) 45(11%)
16(16%) 11 (11%)
100(25%) 118(29%) 86(21%)
27 (28%) 19 (20%) 24 (25%)
A symptoms B symptoms
263 (56%) 205
242 (60%) 160
56 (58%) 41
Stage
9 (2%) 7(1%) 107(23%) 65(14%) 107 (23%) 173(37%)
9 (2%) 7 (2%) 93 (23%) 54(13%) 98 (24%) 141 (35%)
1 (1%) 1 (1%) 18(19%) 6 (6%) 27 (28%) 44 (45%)
96(21%) 244 (52%) 84(18%) 36 (8%) 6(1%)
91 (23%) 208 (52%) 67(17%) 30 (7%) 4(1%)
35 (36%) 32 (33%) 21 (22%) 6 (6%) 1 (1%)
98 (21%)
77(17%)
17(18%)
WEEK 1 2 3 4 5 6 7 (1 9 10 11 12 Doxombichi Cyclophoaphimide Vincristine (no cap)
402 17-74 (median 53) 245:157
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Figure I. Twelve-week combination chemotherapy regimens. All regimens included prednisone, cimetidine, ketoconazole and co-trimoxazole given for 12 weeks. subtypes: diffuse mixed, diffuse large cleave"d, diffuse non-cleaved, diffuse not otherwise specified, or immunoblastic; stages III or IV or stage I or II if associated with B symptoms or a mass greater than 10 cm in maximal size; and age 16-70 years. Patients were ineligible if their lymphoma had arising in the setting of organ transplant or AIDS; had previously been treated; was of discordant histology or had arisen from an antecedant low grade lymphoma or there was any evidence of neoplastic follicle formation i.e., it had solely to be a diffuse process; or if they had significant cardiac impairment. All biopsies were reviewed by an experienced hematopathologist. Six hundred and eighteen eligible patients were seen from 1 April 1981 to 31 December 1995, of whom 468 were treated as above. One hundred and fifty (24%) received other treatments for the following reasons: a less intense protocol designed for the elderly (n = 45); a dose intensification protocol instituted in January 1991 for patients whose risk of relapse or primary progression was estimated to be at least 75% O = 41); too ill to treat (n = 4); a protocol designed for limited stage disease consisting of three cycles of chemotherapy and irradiation (n = 11), and the remaining 49 received other standard lymphoma regimens. Comparison to Provincial Cancer Registry data indicates that 75%80% of patients fulfilling the eligibility criteria and diagnosed within the study period were managed on one of these three 12-week protocols under the auspices of the BC Cancer Agency. Once patients completed the chemotherapy they were classified as non responders (n = 6); partial responders, defined as a less than 50% shrinkage with progression within two months of the end of chemotherapy (n = 52); partial responders who became complete responders using additional irradiation (n = 8); and complete responders (n = 402), defined as a 50% or more shrinkage with no evidence of progression within two months of completing chemotherapy. This cohort of 402 patients form the basis of this report on relapse from 'complete response'. The characteristics at initial diagnosis of the primary cohort of 468, the 402 responders and the 97 who then relapsed are presented in Table 1. The cut-off date for the analysis was 1 July 1996 and all relapses occuring up to this date have been included in this report. Prior to the initiation of the primary chemotherapy all patients were staged with history and physical examination, chest radiograph, CTscan of the abdomen and pelvis, unilateral iliac crest bone marrow biopsy and aspiration; complete blood count; serum lactate dehydro-
ECOG"
1A bulky IB IIA bulky IIB III IV 0 1 2 3 4
Two or more extra nodal sites LDH value" LDH value > normal 0
55-6086 (median 224)
55-6086 (median 219)
259/450(58%) 211/387(55%)
111-3050 (median 200) 51/91(56%)
Abbreviation: NOS - not otherwise specified. * Missing value in two patients. b Upper limit normal = 195. c Denominator less than number of patients due to missing values.
genase (LDH); serum creatinine; serum protein electrophoresis; bilirubin, albumin and calcium. Performance status initially was not routinely recorded (it was retrospectively determined through chart review) but since 1990 has been recorded prospectively. Other investigations were performed as directed by symptoms or known associations between disease sites. All this information was entered on a standard staging diagram at or prior to treatment initiation. Once the primary chemotherapy was completed patients were completely restaged with all abnormal tests repeated. They were then followed but there was no standardized regimen with regard to follow-up frequency or the tests to be performed. However, most were reviewed with history and physical, CBC and LDH at intervals of three months for two years, six months for three years and annually thereafter. Other tests were performed as symptoms indicated. On symptomatic relapse all patients had a history and physical examination but there was no formal policy mandating complete restaging. Our knowledge of the extent of disease at relapse was often less than at diagnosis. The investigations performed at relapse and their frequency are listed in Table 2. Performance status and symptoms were known for the majority. LDH was measured in 74%. All these data were abstracted retrospectively from the patient chart. Tissue at relapse was obtained in 70 (72%): open biopsy (n = 36), fine needle aspirate (n = 26), effusion cytology (n = 1), cerebro-spinal fluid (n = 4), bone marrow (n = 2) and post mortem (n - 1). There was no formal policy with regard to therapy on relapse and the following were used: single agent chemotherapy (n = 12), combina-
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
50mg/mi 35Omg/ro' 50mg/m' 100mg/m' 10mg/m' 1.2mg/ra'
468 17-74 (median 53) 290:178
X
VACOP-B (6/86-4/93)
Doxonibfcin Cycloptosphamtde Etopoiide(dl) Etoposide (d2*3) Bleomycin Vincristine (no cap)
Relapsers from complete response
X X
X
X
Complete responders
X
X
X
Initial cohort
1127 Table 2. Investigations on relapse.
Table 3. Prognostic variables studied in the univariate analysis. No.
Percentage
Variable
Lymphangiogram Bone scan Gallium scan
97 67 69 41 9 9 6 4 3 1
100% 69% 71% 42% 9% 9% 6% 4% 3% 1%
CTabdomen\pelvis/>/u5 CXR CTabdomen\pelvis/>/us CXR plus bone marrow
48 28
49% 29%
< 1 year versus > 1 year M versus f Age at 1st relapse Continuous variable Stage at initial diagnosis I versus II versus III versus IV and as a trend A versus B B symptoms at diagnosis A versus B B symptoms at relapse Performance status at diagnosis 0 versus 1, 0 versus 2, 1 versus 2 and as a trend 0 versus 1, 0 versus 2, 1 versus 2 and Performance status at relapse as a trend Within normal range versus elevated j-ijri at diagnosis Within normal range versus elevated LDH at relapse > 3 versus < 3 Number of nodal sites at diagnosis Number of nodal sites at relapse > 3 versus ^ 3 Of extranodal sites at diagnosis* 0 versus 1 versus 2 and as a trend Number of extranodal sites at 0 versus > / relapse" Histology at relapse Low versus intermediate grade
History and physical CT abdomen and pelvis Chest X-ray Bone marrow biopsy and aspiration CT chest CT brain
CSF
Results
Time to 1st relapse
Sex
° Stage IV patients only.
grade). The follow-up duration of the initial cohort of 402 patients ranged from 0.3-16 years with a median of 4.25 years. Relapse sites were nodal (n = 48, 49%), extranodal (n = 23, 24%) or both (n = 26, 27%). Formal complete restaging was not a prerequisite and so these figures are under estimates. Seventy-six patients (80%) had no B symptoms at relapse. In contrast, at initial diagnosis 58% had no B symptoms. Overall survival andfailure free survival after first relapse (Figures 2 and 3)
Natural history of relapse (n - 97) The relapse rate was 24% (97 of the 402 patients who entered a CR). The late relapse rate (greater than two years from diagnosis) was 7% (29 of 402). Eleven of the relapses (11%) were low grade: ten biopsy proven and one clinically identified. The remainder (n = 85), minus one with missing information, were intermediate grade (61% biopsy proven, 39% clinically diagnosed). The nonbiopsied relapses were assigned either to the low-grade or intermediate grade caterories based upon a subjective interpretation, at the time of the retrospective chart review, of the rate of growth of the relapse. As such, the rate of intermediate grade relapse may be an overestimate (39% not biopsy proven) and converseley the rate of low grade relapse may be an underestimate. Time from initial diagnosis to first relapse is recorded in Table 4. Low grade relapses were seen at all time points with two low grade relapses (18%) within the first year. The latest low grade relapse was in the tenth year. Seventy-five percent of intermediate grade relapses occurred within two years of the original diagnosis. The latest intermediate relapses (n - 2) were seen in the eleventh year. Intermediate grade relapses tend to occur earlier (75% within two years compared to 27% for low
Five-year overall survival (OS) after relapse from a complete response for all 97 patients was 24% with a median survival of 1.2 years. Sixty-four have died from lymphoma, one from toxicity and two from other causes. Table 4. Relapses rate with time by histologic type at relapse. Time (years)
0.5 1.0 1.5 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0
Histologic:type Intermediate grade
Low grade
Number
Cumulative (%)
Number
Cumulative (%)
16 28 11 9 3 3 2 7 2 1 1 0 2
19 52 65 75 79 82 85 93 96 97 98 98 100
0 2 1 0 1 3 1 1 0 1 0 1
0 18 27 27 36 64 73 82 82 91 91 100
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
tion chemotherapy (n = 36), high dose chemotherapy/radiotherapy and hematopoietic stem cell transplantation (n = 11), radiotherapy (n = 11), chemotherapy and radiotherapy (n = 15), surgery (n = 2), chemotherapy and surgery (n = 2), and no therapy (n = 8) Overall survival was recorded from the date of first relapse until last follow up, or death from any cause Status was recorded as alive with no evidence of disease, alive with lymphoma, dead due to lymphoma, dead due to toxicity and dead of other causes (i.e., not treatment-related toxicity or lymphoma). Failure free survival from relapse classifies as failure any of the following: second relapse, progression, death from lymphoma in the absence of documented second relapse, or toxic death and was recorded from the date of first relapse. The method of Kaplan-Meier was used to plot survival. Univanate analysis of the potential prognostic factors listed in Table 3 was carried out using log-rank statistics. Multivariate analysis for independent factors was then carried out using those factors significant by univariate analysis [14].
Comparison
1128 Table 5. Outcome by treatment at first relapse. A)
Intermediate grade (n = 85)*
NED AWD
Toxic death
Dead other
2 3 4 6 3
2 1 1
1
9 29 5 6 3 1 7
1 1
Year Figure 2. Overall survival from first relapse. all patients (n = 97), — intermediate grade relapse (n = 85); low grade relapse (n = 11).
B) Low grade (/i = 11) NED
AWD
100rr
Dead lymphoma
Toxic death
Dead other
60-
Single agent chemotherapy Combination chemotherapy Chemotherapy/ irradiation Irradiation Surgery
20
Abbreviations: OS - overall survival; NED - alive and no evidence of lymphoma; AWD - alive with lymphoma. * One patient with missing data on histology. b Eight autologous marrow; three allogeneic.
s«
a o
8 Year Figure 3. Failure free survival from first relapse. 97); - - - intermediate grade relapse (n = 85); (/i = 1 1 ) .
10
all patients (n = low grade relapse
1 1 2 1
2 1
1
1 1
OS and FFS: time from diagnosis to first relapse, performance status at relapse, B symptoms at relapse, LDH at relapse, number of nodal sites at relapse, and number of extranodal sites at original diagnosis. In addition, for FFS, number of extranodal sites at relapse attained significance. None of the other comparisons were statistically significant. OS for each of the significant factors is presented graphically in Figures 4 and 5.
Failure free survival (FFS) was 18% at five years with a median of 0.6 years. OS and FFS for low-grade versus intermediate grade histology at relapse are illustrated in Multivariate analysis Figures 2 and 3. Eighteen patients with intermediate grade relapse are alive in maintained second remission The multivariate analysis was performed on the 85 with follow up of 1.5-11 years. patients with intermediate grade relapse using all variables identified as significant in the univariate analysis. Only patients with data for all of these variables could be Outcome by treatment given at first relapse (Table 5) included. Therefore, as the LDH value was only availTreatment was at the treating physician's discretion able in 63 of 85 the analysis was performed both including and excluding LDH. Independent factors, with LDH hence the wide variety of options. values excluded, for OS were: duration from initial diagnosis to relapse (P - 0.005), performance status at Univariate survival analysis (Table 6) relapse (P - 0.002), and number of involved nodes at Histologic subtype at relapse (low-grade versus inter- relapse (P - 0.03); including LDH, then the following mediate grade) was a significant factor in the univariate were independent factors for OS: duration from initial analysis (P = 0.037) with longer overall survival for diagnosis to relapse (P = 0.002), performance status at those patients with low-grade relapse (Figure 2). In light relapse (P = 0.005), and number of nodal sites at relapse of the different natural histories of these histologic (P = 0.04). LDH was not significant (P = 0.6). For FFS, groups, the rest of the univariate analysis was performed excluding LDH, the following were independently sigonly on those with intermediate grade histology at nificant: time to first relapse {P = 0.008), performance relapse {n = 85). The following factors from those listed status at relapse (P = 0.02), and number of nodal sites at in Table 3 reached statistical significance (P ^ 0.05) for relapse (P - 0.02). Again including LDH did not alter
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
Chemotherapy/ surgery Single agent chemotherapy Combination chemotherapy Marrow transplant15 Chemotherapy/ irradiation Irradiation Surgery No treatment
Dead lymphoma
1129 Table 6. Univariate analysis of factors affecting survival in patients with intermediate grade relapse (n = 85). Factor
Time to 1st relapse Performance status 0
< 1 year 3+2 versus 1 versus 0 1 (versus 0) 3+2 (versus 0) B
> normal >3 > 1 versus 1 versus 0 > 1 versus 1 versus 0
Overall survival*'b
Failure free survival* lb
i>-value
P-value
Median years
0.0004 0.0002 0.003 0.0001 0.01 0.001 0.046 0.001
0.5 versus 1.0
83 64 82 83
0.0009 0.6 versus 2.2 0.0002 0.003 0.8 versus 2.4 0.0001 0.2 versus 2.4 0.002 0.45 versus 1.3 0.002 0.9 versus 2.3 0.02 0.8 versus 1.2 Not significant
83
0.05
0.05
No. of patients
85 82
Median years
0.4 versus 0.9 versus 1
0.6 versus 1.3 0.2 versus 1.3 0.30 versus 1.2 0.4 versus 1.0 0.5 versus 0.9 0.5 versus 0.9 versus 0.9 0.25 versus 0.8 versus 0.8
" Measured from date of relapse. b The lowest median survival is that of the adverse factor. The other values relate to the comparator (see Table 3). 0 At relapse. d At initial diagnosis.
these results and LDH itself was not independently significant. Discussion This retrospective study was performed to discover the temporal sequence and histology of relapse, the outcome after relapse and prognostic factors predictive of outcome in a group of patients who had been uniformly treated at initial diagnosis with multi-agent chemotherapy and who entered a complete remission. Four hundred sixty-eight patients aged 17-74 years with advanced diffuse stage large cell or immunoblastic lymphoma have, since April 1981, been treated on sequentially developed 12-week chemotherapy regimens (13). Of these, 402 (86%) entered a complete response and 97 (24%) of these patients subsequently relapsed with the latest relapse in the eleventh year after diagnosis. Eleven of these first relapses (11%) were of low grade histology (i.e., a discordant relapse) by the Working Formulation [15]. A caveat is needed. No biopsy was performed in 28% of all the relapsing patients and this frequency of low grade relapse may, therefore, be an underestimate. Such low-grade relapses have been reported in the past, both as case reports (16-21) and as larger studies [7, 8]. The MD Anderson report included only late relapses (> 30 months from the start of chemotherapy) and included patients who had discordant histology (i.e., both low and intermediate grade at separate sites) at initial diagnosis [7]. In contrast, patients with either an antecedant low grade lymphoma or known discordant histology at primary diagnosis were not included in our study cohort. In the Dana-Farber report two of their eleven late relapses (>24 months post treatment) were low-grade. The 11% of relapses with low-grade histology noted in our series may represent the de novo occurrence of a second lymphoma or the low grade histology may have been present at diagnosis and
missed. Consistent with this latter hypothesis is that 18% of patients who had multiple biopsies (sites and number of sites were not specified) at the National Cancer Institute between 1953 and 1975 were recognized as having discordant histologies [22]. Late relapse of diffuse large cell lymphoma (more than two years from diagnosis) is believed by most oncologists to be an an unusual occurrence. This is an erroneous belief. Twenty-five per cent of the intermediate grade relapses occurred more than two years after the initial diagnosis with the latest being in the eleventh year. This is similar to the Dana-Farber Experience with M-BACOD in which 32% of the relapses, including low grade, occurred after two years [8]. Overall our late relapse rate was 7% (29 of 402) which is similar to that in other series [7, 8]. This will be an under estimate of the true rate of late relapses, given that they occured up to 11 years from diagnosis, as our median follow up was only 4.25 years (range 0.3-16 years). Late relapses may be due to de novo lymphoma arising in a susceptible host with pre-existing chromosomal abnormalities or it may represent recurrence of the original lymphoma. Three patients in the M.D. Anderson series (relapsing at 2.5-4 years) had tissue from diagnosis and relapse available for comparison. Phenotype and DNA content were the same implying recurrence [7]. In an earlier report of one case from our group Southern blot analysis of immunoglobulin heavy chain rearrangement demonstrated an identical pattern for the relapse (at 57 months) and the original tumour [23], Contrary to common belief, death from lymphoma, after relapse, is not inevitable. Overall survival after relapse was a median of 1.2 years with afiveyear rate of 24%. For failure free survival the corresponding figures were 0.6 years and 18%. For those patients with intermediate grade relapse the median overall survival was one year with a 20% five year rate; for failure free survival 0.6 years and 17%. Long term survival is therefore possible and appropriate therapy for relapsed inter-
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
Symptoms0 LDHC Number of nodal sites0 Number of extranodal sites0 Number of extranodaJ sites'1
Adverse factor
1130 a)
100=1
p = <0 001
p=00O2
20-
S ^ - - ^ * . . , . . . ,
4
6
10
b)
100 ?
80 p = 0001
C) 1
10
0
2
4
6
8
10
r
Figure 4. Overall survival for intermediate grade relapses from relapse. a) LDH value at relapse $ upper limit normal (n = 32) > upper limit normal (n = 32) b) Number of extranodal sites at primary diagnosis None (TI = 44) One(7i = 23) More than one (TI = 16) c) Symptoms at relapse A symptoms (TI = 65) B symptoms (TI = 18)
first
mediate grade disease should be instituted. High dose chemotherapy with marrow transplantation has been shown to be the treatment of choice in selected younger patients with chemosensitive disease (11). The results of the Parma trial with, after response to initial DHAP therapy, a randomization between either high dose chemotherapy/autologous bone marrow transplantation or further chemotherapy with DHAP have given further
Figure 5. Overall survival from first relapse. ) Performance status (ECOG) 0 (n = 30) 1 (n = 32) 2 (n = 14) 3 (n = 6) b) Time from initial diagnosis to first relapse < 1 year (n = 44) > 1 year (TI = 40) c) Number of nodal sites at relapse < 3 (n = 62) > 3 (TI = 20)
a
support to this approach [24]. However, a caveat is needed as evidenced by the outcomes reported in this paper. The relapse therapies chosen for the patients in this report were selected by their physicians based upon their perception of the patients likely outcome. Six of the twelve who had apparently localized relapse received combination chemotherapy and involved field irradiation and are still alive in remission. Similarly three of
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
Year
1131
In conclusion, 24% of non-elderly patients treated with 12-week combination chemotherapy for advanced diffuse large-cell lymphoma who entered a complete response relapsed with the majority of relapses within two years of original diagnosis. However, 25% of the relapses were delayed with the latest recorded in the eleventh year. Eleven percent of these first relapses were of low-grade histology which has therapeutic implications and demonstrates the value of biopsy verification. Independent prognostic variables were identified at relapse predictive of outcome. These need to be validated and then, if reproducible, accounted for when analyzing results of salvage therapy.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 15.
16.
Acknowledgements
Jan Wilson for her secretarial help and the physicians of the Province of British Columbia who referred the patients.
References 1. Weick JK, Dahlberg S, Fisher RI et al. Combination chemotherapy of intermediate-grade and high-grade non-Hodgkin's lymphoma with MACOP-B: A Southwest Oncology Group Study. J Clin Oncol 1991; 9: 748-53. 2. Longo DT, Devita VT, Duffey PL et al. Superiority of ProMACECytaBOM over ProMACE-MOPP in the treatment of advanced diffuse aggressive lymphoma: Results of a prospective randomized trial. J Clin Oncol 1991; 9: 25-38. 3. Dana BW, Dahlberg S, Miller TP et al. m-BACOD treatment for intermediate and high grade malignant lymphomas; A South-
17.
18.
19.
20.
21.
22.
west Oncology Group phase II trial. J Clin Oncol 1990; 8: 1155— 62. Fisher RI, Miller TP, Dana BW et al. Southwest Oncology Group clinical trials for intermediate and high grade non-Hodgkin's lymphoma. Semin Haematol 1987; 24: 21-5. Fisher RI, Gaynor ER, Dahlberg W et al. Comparison of a standard regimen (CHOP) with three intensive chemotherapy regimens for advanced non-Hodgkin's lymphoma. N Engl J Med 1993; 328: 1002-6. Coiffier B, Gisselbrecht C, Herbrecht R et al. LNH-84 Regimen: A multicenter study of intensive chemotherapy in 737 patients with aggressive malignant lymphoma. J Clin Oncol 19889; 7: 1018-26. Cabanillas F, Velasquez WS, Hagenmeister FB et al. Clinical, biological and histologic features of late relapses in diffuse large cell lymphoma. Blood 1992; 79: 1024-8. Shipp MA, Yeap BY, Harrington DP et al. The m-BACOD combination chemotherapy regimen in large-cell lymphoma. Analysis of the complete trial and comparison with the M-BACOD regimen. J Clin Oncol 1990; 8: 84-93. Coltman CA, Dahlberg S, Iches SE et al. CHOP is curative in thirty percent of patients with large cell lymphoma. A twelve year Southwest Oncology Group follow-up. In Skarin AT (ed): Update on Treatment for Diffuse Large Cell Lymphoma. New York: Park Row 1986; 71-81. Philip T, Hagenbeek A, Gomez F et al. Salvage treatment for chemotherapy resistant diffuse non-Hodgkin's lymphoma (NHL). Review of the 215 patients of the Parma study and rationale for a second prospective study. Ann Oncol 1986; 7 (Suppl 3): 17 (Abstr 54). Bosly A, Coiffier B, Gisselbrecht C et al. Bone marrow transplantation prolongs survival after relapse in aggressive lymphoma patients treated with the LNH-84 regimen. J Clin Oncol 1992; 10: 1615-23. Blay J-Y, Gomez F, Bachelot Tet al. The international prognostic index correlates to survival in intermediate/high grade lymphoma (NHL) in relapse: Analysis of the Parma trial. Proc Am Soc Clin Oncol 1996; 15: 151 (Abstr 269). Hoskins P, Connors JM, Gascoyne R et al. 10-year outcome of 12-week duration chemotherapy for advanced large-cell lymphoma: The Vancouver experience. Ann Oncol 1996; 7: 58 (Abstr 198). Kalbfleisch JD, PTentice RL. The Statistical Analysis of Failure Time Data. New York: Wiley 1980. Non-Hodgkin's Lymphoma Pathologic Classification Project: National Cancer Institute sponsored study of classifications of non-Hodgkin's lymphoma. Summary and description of a working formulation for clinical usage. Cancer 1982; 49: 2112-35. Marazuela M, Yebra M, Giron JA et al. Late relapse with nodular lymphoma after treatment for diffuse non-Hodgkin's lymphoma. Cancer 1991; 67: 1950-3. Siegel RD, Skarin AT, Canellos GP. Relapse with nodular lymphoma following combination chemotherapy for diffuse nonHodgkin's lymphoma. Am J Clin Oncol 1986; 9: 420-3. O'Brien CJ, Howard M, Child JA. Downgrading of non-Hodgkin's lymphoma following chemotherapy. Acta Haematol 1987; 77: 57-9. Head DR, Avakian J, Kjeklsberg R, Cerezo L. Relapse of intermediate or high-grade (unfavorable) non-Hodgkin's lymphoma as a low-grade (favorable) non-Hodgkin's lymphoma. Am J Clin Pathol 1988; 289: 106-8. Cheragan DP, Four K, Dressier L. High grade non-Hodgkin's lymphoma relapsing as low-grade follicular lymphoma. So-called downgraded lymphoma. Am J Hematol 1989; 30: 36-4. Mead GM, Mackintosh FR, Burke JS, Rosenberg SA. Late relapse from complete remission in nodular and diffuse histiocytic lymphoma. Cancer 1983; 52: 1356-9. Fisher RI, Jones RB, DeVita VT et al. Natural history of malignant lymphomas with divergent histologies at staging evaluation. Cancer 1981; 47: 2022-5.
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
eight treated with irradiation alone are in continuing remission. Patient selection alone, regardless of the treatment used, may alter outcome. It is therefore essential that in randomized trials the patient arms be equivalent with regard to their prognostic mix. Therefore, in an attempt to identify which factors identifiable at relapse most influenced outcome we performed univariate and multivariate analysis in the patients with intermediate grade relapse. Adverse independent factors both for overall survival and failure free survival were: a duration from initial diagnosis to first relapse of less than one year, decreasing performance status at relapse, and more than three nodal sites at relapse. These same factors were important whether or not the LDH value was included in the analysis. LDH was not an independent factor but it was only measured in 74% and so a true effect may have been missed. Limitations of this analysis were that uniform treatment on relapse was not used, that the extent of re-staging varied, and that elderly patients were not part of the initial treatment cohort. As such these results should only be regarded as hypothesis generating. The prognostic factor analysis does need to be repeated in data sets where the above limitations do not exist. Despite this the results are consistent with those from other such analyses performed at initial diagnosis [25] and at relapse [7,10-12].
1132 23. Lee AYY, Connors JM, Klimo P et al. Late relapse in patients with diffuse large cell lymphoma treated with MACOP-B. J Clin Oncol 1997; 15: 1445-1753. 24. Philip T, Guglielmi C, Hagenbeelc A et al. Autologous bone marrow transplantation as compared with salvage chemotherapy in relapses of chemotherapy-sensitive non-Hodgkin's lymphoma. N Engl J Med 1995; 333: 1540-5. 25. The International Non-Hodgltin's lymphoma Prognostic Factors Project. A predictive model for aggressive non-Hodgkin's lymphoma. N Engl J Med 1993; 329: 987-94.
Received 20 June 1997; accepted 25 August 1997.
Correspondence to: Dr. P. J. Hoskins British Columbia Cancer Agency Div. of Medical Oncology 600 West 10th Avenue Vancouver British Columbia, Canada V5Z4E6
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
Original article Advanced diffuse large-cell lymphoma treated with 12-week combination chemotherapy: Natural history of relapse after initial complete response and prognostic variables defining outcome after relapse P. I Hoskins,1 N. Le,2 R. D. Gascoyne,3 R. Klasa,1 T. Shenkier,1 S. O'Reilly1 & J. M. Connors1
Summary Purpose: To define both the natural history of and prognostic factors affecting outcome post relapse from a complete response in advanced stage diffuse large-cell lymphoma. Patients and methods: A total of 468 patients aged 17-74 years received the 12-week duration chemotherapy regimens MACOP-B.VACOP-B and ACOP-12 between 1 April 1981 and 31 December 1995 for advanced stage diffuse large, mixed or immunoblastic lymphoma. Of these 402 entered a complete remission, 97 (24%) of whom subsequently relapsed. Initial staging data, follow-up, and relapse information were analyzed to define the natural history of relapse and also subjected to univariate and multivariate correlation with overall (OS) and failure free survival (FFS). Results: Eleven percent of the relapses were low grade. All other relapses were of intermediate grade with 75% occurring
Introduction
within the first two years, the remainder up until the eleventh year. Median and five-year OS from the time of relapse for intermediate grade relapse were 12 months and 20%; for FFS they were eight months and 18% respectively. Adverse independent factors, for both OS and FFS were: less than one year to relapse, decreasing performance status at relapse, and more than three nodal sites at relapse. Conclusions: Low-grade relapse is not uncommon in patients who initially presented with diffuse large cell lymphoma. As the management of low- and intermediate grade disease is so different biopsy proof of the nature of the relapse is of value. The prognostic factors identified need to be taken into consideration when analyzing results from trials of secondary treatment so as to avoid erroneous conclusions about comparative treatment efficacy. Key words: lymphoma, relapse
tumour mass, high LDH level and non low-grade histology [7, 11, 12]. The relevance of this is that if these factors are not stratified for they could lead to survival differences in salvage trials which may falsely be attributed to treatment. In order to further elucidate the natural history of and prognostic factors related to survival after relapse, we conducted this retrospective study of patients with advanced, diffuse large-cell lymphoma who have all been treated with 12-week duration combination chemotherapy.
Up to 40% of patients with advanced diffuse large-cell or immunoblastic lymphoma who respond to first-line multi-agent chemotherapy relapse [1-6]. It is commonly believed that most, if not all, relapses occur within the first 24 months but there are reports of later relapses. In the M.D. Anderson experience 5% of the relapses occurred more than 30 months from diagnosis, with the longest interval being 170 months [7]. The group from the Dana-Farber found a 7.8% late relapse rate, occurring up to seven years [8]. In both of these series lowgrade (i.e., discordant) relapses were also seen. In the SWOG series using CHOP as the primary chemotherPatients and methods apy relapses were seen up to seven years from diagnosis [9]. These later relapses may be biologically different Twelve-week duration chemotherapy (MACOP-B April 1981 to May from early relapses. There is indirect evidence for such a 1986; VACOP-B June 1986 to April 1993 and ACOP12 May 1993 and biologic difference from clinical data. In the Parma continuing) has been our standard treatment for advanced large cell randomized study of salvage chemotherapies the two- lymphoma since 1981 (Figure 1). These three regimens are based on a year survival rate was 19% if the first relapse occurred backbone of doxorubicin, cyclophosphamide, vincristine and predniThey were equivalent with regard to survival outcomes when within 12 months of the initial diagnosis versus 47% if it sone. analyzed within the subgroupings of the International Prognostic Facwere after 12 months [10]. Other prognostic factors have tors Index and so have been considered as one uniform group for the been identified at relapse as predicting for worse sur- purposes of this paper [13]. vival and these include poor performance status, large The eligibility criteria to receive this chemotherapy were: histologic
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
^Division of Medical Oncology, 2Division of Epidemiology, 3Department of Pathology. British Columbia Cancer Agency, Vancouver Centre and the University of British Columbia, Vancouver, British Columbia
1126 MACOP-B (4/81-5/86)
Table 1. Patient characteristics at initial diagnosis. WEEK 1 2 3 4 5 6 7 8 9 10 11 12
Doxorubicm Cyclophojphmude Vmcnfline (ctp-2.0mg) Blcoraycin Mcthotrexate
50meW 35Omg/m' 1 4 mg/m' 10mg/ra' 400mg/mJ
rv IV
X
X
X
X
X
X
X
X
X
X
rv
X
X
X
IV
X
rv
X X
Number Age (years) Male: female
WEEK 1 2 3 4 5 6 7 8 9 10 11 12
rv
IV IV PO IV IV
X
X
X
X
X
X
X
X
X
X X
X
X X X
X
X
X
X X X
X
X
X
X
X
X
ACOP-12 (5/93 - and continuing)
50mg/mJ 35Omg/ra' 1.2mg/m'
IV IV IV
97 21-68 (median 55) 51:46
Diffuse mixed 61(13%) Diffuse large-cleaved 52(11%) Diffuse large non-cleaved 114(24%) Diffuse large NOS 134(29%) Immunoblastic 107 (23%)
53(13%) 45(11%)
16(16%) 11 (11%)
100(25%) 118(29%) 86(21%)
27 (28%) 19 (20%) 24 (25%)
A symptoms B symptoms
263 (56%) 205
242 (60%) 160
56 (58%) 41
Stage
9 (2%) 7(1%) 107(23%) 65(14%) 107 (23%) 173(37%)
9 (2%) 7 (2%) 93 (23%) 54(13%) 98 (24%) 141 (35%)
1 (1%) 1 (1%) 18(19%) 6 (6%) 27 (28%) 44 (45%)
96(21%) 244 (52%) 84(18%) 36 (8%) 6(1%)
91 (23%) 208 (52%) 67(17%) 30 (7%) 4(1%)
35 (36%) 32 (33%) 21 (22%) 6 (6%) 1 (1%)
98 (21%)
77(17%)
17(18%)
WEEK 1 2 3 4 5 6 7 (1 9 10 11 12 Doxombichi Cyclophoaphimide Vincristine (no cap)
402 17-74 (median 53) 245:157
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Figure I. Twelve-week combination chemotherapy regimens. All regimens included prednisone, cimetidine, ketoconazole and co-trimoxazole given for 12 weeks. subtypes: diffuse mixed, diffuse large cleave"d, diffuse non-cleaved, diffuse not otherwise specified, or immunoblastic; stages III or IV or stage I or II if associated with B symptoms or a mass greater than 10 cm in maximal size; and age 16-70 years. Patients were ineligible if their lymphoma had arising in the setting of organ transplant or AIDS; had previously been treated; was of discordant histology or had arisen from an antecedant low grade lymphoma or there was any evidence of neoplastic follicle formation i.e., it had solely to be a diffuse process; or if they had significant cardiac impairment. All biopsies were reviewed by an experienced hematopathologist. Six hundred and eighteen eligible patients were seen from 1 April 1981 to 31 December 1995, of whom 468 were treated as above. One hundred and fifty (24%) received other treatments for the following reasons: a less intense protocol designed for the elderly (n = 45); a dose intensification protocol instituted in January 1991 for patients whose risk of relapse or primary progression was estimated to be at least 75% O = 41); too ill to treat (n = 4); a protocol designed for limited stage disease consisting of three cycles of chemotherapy and irradiation (n = 11), and the remaining 49 received other standard lymphoma regimens. Comparison to Provincial Cancer Registry data indicates that 75%80% of patients fulfilling the eligibility criteria and diagnosed within the study period were managed on one of these three 12-week protocols under the auspices of the BC Cancer Agency. Once patients completed the chemotherapy they were classified as non responders (n = 6); partial responders, defined as a less than 50% shrinkage with progression within two months of the end of chemotherapy (n = 52); partial responders who became complete responders using additional irradiation (n = 8); and complete responders (n = 402), defined as a 50% or more shrinkage with no evidence of progression within two months of completing chemotherapy. This cohort of 402 patients form the basis of this report on relapse from 'complete response'. The characteristics at initial diagnosis of the primary cohort of 468, the 402 responders and the 97 who then relapsed are presented in Table 1. The cut-off date for the analysis was 1 July 1996 and all relapses occuring up to this date have been included in this report. Prior to the initiation of the primary chemotherapy all patients were staged with history and physical examination, chest radiograph, CTscan of the abdomen and pelvis, unilateral iliac crest bone marrow biopsy and aspiration; complete blood count; serum lactate dehydro-
ECOG"
1A bulky IB IIA bulky IIB III IV 0 1 2 3 4
Two or more extra nodal sites LDH value" LDH value > normal 0
55-6086 (median 224)
55-6086 (median 219)
259/450(58%) 211/387(55%)
111-3050 (median 200) 51/91(56%)
Abbreviation: NOS - not otherwise specified. * Missing value in two patients. b Upper limit normal = 195. c Denominator less than number of patients due to missing values.
genase (LDH); serum creatinine; serum protein electrophoresis; bilirubin, albumin and calcium. Performance status initially was not routinely recorded (it was retrospectively determined through chart review) but since 1990 has been recorded prospectively. Other investigations were performed as directed by symptoms or known associations between disease sites. All this information was entered on a standard staging diagram at or prior to treatment initiation. Once the primary chemotherapy was completed patients were completely restaged with all abnormal tests repeated. They were then followed but there was no standardized regimen with regard to follow-up frequency or the tests to be performed. However, most were reviewed with history and physical, CBC and LDH at intervals of three months for two years, six months for three years and annually thereafter. Other tests were performed as symptoms indicated. On symptomatic relapse all patients had a history and physical examination but there was no formal policy mandating complete restaging. Our knowledge of the extent of disease at relapse was often less than at diagnosis. The investigations performed at relapse and their frequency are listed in Table 2. Performance status and symptoms were known for the majority. LDH was measured in 74%. All these data were abstracted retrospectively from the patient chart. Tissue at relapse was obtained in 70 (72%): open biopsy (n = 36), fine needle aspirate (n = 26), effusion cytology (n = 1), cerebro-spinal fluid (n = 4), bone marrow (n = 2) and post mortem (n - 1). There was no formal policy with regard to therapy on relapse and the following were used: single agent chemotherapy (n = 12), combina-
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
50mg/mi 35Omg/ro' 50mg/m' 100mg/m' 10mg/m' 1.2mg/ra'
468 17-74 (median 53) 290:178
X
VACOP-B (6/86-4/93)
Doxonibfcin Cycloptosphamtde Etopoiide(dl) Etoposide (d2*3) Bleomycin Vincristine (no cap)
Relapsers from complete response
X X
X
X
Complete responders
X
X
X
Initial cohort
1127 Table 2. Investigations on relapse.
Table 3. Prognostic variables studied in the univariate analysis. No.
Percentage
Variable
Lymphangiogram Bone scan Gallium scan
97 67 69 41 9 9 6 4 3 1
100% 69% 71% 42% 9% 9% 6% 4% 3% 1%
CTabdomen\pelvis/>/u5 CXR CTabdomen\pelvis/>/us CXR plus bone marrow
48 28
49% 29%
< 1 year versus > 1 year M versus f Age at 1st relapse Continuous variable Stage at initial diagnosis I versus II versus III versus IV and as a trend A versus B B symptoms at diagnosis A versus B B symptoms at relapse Performance status at diagnosis 0 versus 1, 0 versus 2, 1 versus 2 and as a trend 0 versus 1, 0 versus 2, 1 versus 2 and Performance status at relapse as a trend Within normal range versus elevated j-ijri at diagnosis Within normal range versus elevated LDH at relapse > 3 versus < 3 Number of nodal sites at diagnosis Number of nodal sites at relapse > 3 versus ^ 3 Of extranodal sites at diagnosis* 0 versus 1 versus 2 and as a trend Number of extranodal sites at 0 versus > / relapse" Histology at relapse Low versus intermediate grade
History and physical CT abdomen and pelvis Chest X-ray Bone marrow biopsy and aspiration CT chest CT brain
CSF
Results
Time to 1st relapse
Sex
° Stage IV patients only.
grade). The follow-up duration of the initial cohort of 402 patients ranged from 0.3-16 years with a median of 4.25 years. Relapse sites were nodal (n = 48, 49%), extranodal (n = 23, 24%) or both (n = 26, 27%). Formal complete restaging was not a prerequisite and so these figures are under estimates. Seventy-six patients (80%) had no B symptoms at relapse. In contrast, at initial diagnosis 58% had no B symptoms. Overall survival andfailure free survival after first relapse (Figures 2 and 3)
Natural history of relapse (n - 97) The relapse rate was 24% (97 of the 402 patients who entered a CR). The late relapse rate (greater than two years from diagnosis) was 7% (29 of 402). Eleven of the relapses (11%) were low grade: ten biopsy proven and one clinically identified. The remainder (n = 85), minus one with missing information, were intermediate grade (61% biopsy proven, 39% clinically diagnosed). The nonbiopsied relapses were assigned either to the low-grade or intermediate grade caterories based upon a subjective interpretation, at the time of the retrospective chart review, of the rate of growth of the relapse. As such, the rate of intermediate grade relapse may be an overestimate (39% not biopsy proven) and converseley the rate of low grade relapse may be an underestimate. Time from initial diagnosis to first relapse is recorded in Table 4. Low grade relapses were seen at all time points with two low grade relapses (18%) within the first year. The latest low grade relapse was in the tenth year. Seventy-five percent of intermediate grade relapses occurred within two years of the original diagnosis. The latest intermediate relapses (n - 2) were seen in the eleventh year. Intermediate grade relapses tend to occur earlier (75% within two years compared to 27% for low
Five-year overall survival (OS) after relapse from a complete response for all 97 patients was 24% with a median survival of 1.2 years. Sixty-four have died from lymphoma, one from toxicity and two from other causes. Table 4. Relapses rate with time by histologic type at relapse. Time (years)
0.5 1.0 1.5 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0
Histologic:type Intermediate grade
Low grade
Number
Cumulative (%)
Number
Cumulative (%)
16 28 11 9 3 3 2 7 2 1 1 0 2
19 52 65 75 79 82 85 93 96 97 98 98 100
0 2 1 0 1 3 1 1 0 1 0 1
0 18 27 27 36 64 73 82 82 91 91 100
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
tion chemotherapy (n = 36), high dose chemotherapy/radiotherapy and hematopoietic stem cell transplantation (n = 11), radiotherapy (n = 11), chemotherapy and radiotherapy (n = 15), surgery (n = 2), chemotherapy and surgery (n = 2), and no therapy (n = 8) Overall survival was recorded from the date of first relapse until last follow up, or death from any cause Status was recorded as alive with no evidence of disease, alive with lymphoma, dead due to lymphoma, dead due to toxicity and dead of other causes (i.e., not treatment-related toxicity or lymphoma). Failure free survival from relapse classifies as failure any of the following: second relapse, progression, death from lymphoma in the absence of documented second relapse, or toxic death and was recorded from the date of first relapse. The method of Kaplan-Meier was used to plot survival. Univanate analysis of the potential prognostic factors listed in Table 3 was carried out using log-rank statistics. Multivariate analysis for independent factors was then carried out using those factors significant by univariate analysis [14].
Comparison
1128 Table 5. Outcome by treatment at first relapse. A)
Intermediate grade (n = 85)*
NED AWD
Toxic death
Dead other
2 3 4 6 3
2 1 1
1
9 29 5 6 3 1 7
1 1
Year Figure 2. Overall survival from first relapse. all patients (n = 97), — intermediate grade relapse (n = 85); low grade relapse (n = 11).
B) Low grade (/i = 11) NED
AWD
100rr
Dead lymphoma
Toxic death
Dead other
60-
Single agent chemotherapy Combination chemotherapy Chemotherapy/ irradiation Irradiation Surgery
20
Abbreviations: OS - overall survival; NED - alive and no evidence of lymphoma; AWD - alive with lymphoma. * One patient with missing data on histology. b Eight autologous marrow; three allogeneic.
s«
a o
8 Year Figure 3. Failure free survival from first relapse. 97); - - - intermediate grade relapse (n = 85); (/i = 1 1 ) .
10
all patients (n = low grade relapse
1 1 2 1
2 1
1
1 1
OS and FFS: time from diagnosis to first relapse, performance status at relapse, B symptoms at relapse, LDH at relapse, number of nodal sites at relapse, and number of extranodal sites at original diagnosis. In addition, for FFS, number of extranodal sites at relapse attained significance. None of the other comparisons were statistically significant. OS for each of the significant factors is presented graphically in Figures 4 and 5.
Failure free survival (FFS) was 18% at five years with a median of 0.6 years. OS and FFS for low-grade versus intermediate grade histology at relapse are illustrated in Multivariate analysis Figures 2 and 3. Eighteen patients with intermediate grade relapse are alive in maintained second remission The multivariate analysis was performed on the 85 with follow up of 1.5-11 years. patients with intermediate grade relapse using all variables identified as significant in the univariate analysis. Only patients with data for all of these variables could be Outcome by treatment given at first relapse (Table 5) included. Therefore, as the LDH value was only availTreatment was at the treating physician's discretion able in 63 of 85 the analysis was performed both including and excluding LDH. Independent factors, with LDH hence the wide variety of options. values excluded, for OS were: duration from initial diagnosis to relapse (P - 0.005), performance status at Univariate survival analysis (Table 6) relapse (P - 0.002), and number of involved nodes at Histologic subtype at relapse (low-grade versus inter- relapse (P - 0.03); including LDH, then the following mediate grade) was a significant factor in the univariate were independent factors for OS: duration from initial analysis (P = 0.037) with longer overall survival for diagnosis to relapse (P = 0.002), performance status at those patients with low-grade relapse (Figure 2). In light relapse (P = 0.005), and number of nodal sites at relapse of the different natural histories of these histologic (P = 0.04). LDH was not significant (P = 0.6). For FFS, groups, the rest of the univariate analysis was performed excluding LDH, the following were independently sigonly on those with intermediate grade histology at nificant: time to first relapse {P = 0.008), performance relapse {n = 85). The following factors from those listed status at relapse (P = 0.02), and number of nodal sites at in Table 3 reached statistical significance (P ^ 0.05) for relapse (P - 0.02). Again including LDH did not alter
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
Chemotherapy/ surgery Single agent chemotherapy Combination chemotherapy Marrow transplant15 Chemotherapy/ irradiation Irradiation Surgery No treatment
Dead lymphoma
1129 Table 6. Univariate analysis of factors affecting survival in patients with intermediate grade relapse (n = 85). Factor
Time to 1st relapse Performance status 0
< 1 year 3+2 versus 1 versus 0 1 (versus 0) 3+2 (versus 0) B
> normal >3 > 1 versus 1 versus 0 > 1 versus 1 versus 0
Overall survival*'b
Failure free survival* lb
i>-value
P-value
Median years
0.0004 0.0002 0.003 0.0001 0.01 0.001 0.046 0.001
0.5 versus 1.0
83 64 82 83
0.0009 0.6 versus 2.2 0.0002 0.003 0.8 versus 2.4 0.0001 0.2 versus 2.4 0.002 0.45 versus 1.3 0.002 0.9 versus 2.3 0.02 0.8 versus 1.2 Not significant
83
0.05
0.05
No. of patients
85 82
Median years
0.4 versus 0.9 versus 1
0.6 versus 1.3 0.2 versus 1.3 0.30 versus 1.2 0.4 versus 1.0 0.5 versus 0.9 0.5 versus 0.9 versus 0.9 0.25 versus 0.8 versus 0.8
" Measured from date of relapse. b The lowest median survival is that of the adverse factor. The other values relate to the comparator (see Table 3). 0 At relapse. d At initial diagnosis.
these results and LDH itself was not independently significant. Discussion This retrospective study was performed to discover the temporal sequence and histology of relapse, the outcome after relapse and prognostic factors predictive of outcome in a group of patients who had been uniformly treated at initial diagnosis with multi-agent chemotherapy and who entered a complete remission. Four hundred sixty-eight patients aged 17-74 years with advanced diffuse stage large cell or immunoblastic lymphoma have, since April 1981, been treated on sequentially developed 12-week chemotherapy regimens (13). Of these, 402 (86%) entered a complete response and 97 (24%) of these patients subsequently relapsed with the latest relapse in the eleventh year after diagnosis. Eleven of these first relapses (11%) were of low grade histology (i.e., a discordant relapse) by the Working Formulation [15]. A caveat is needed. No biopsy was performed in 28% of all the relapsing patients and this frequency of low grade relapse may, therefore, be an underestimate. Such low-grade relapses have been reported in the past, both as case reports (16-21) and as larger studies [7, 8]. The MD Anderson report included only late relapses (> 30 months from the start of chemotherapy) and included patients who had discordant histology (i.e., both low and intermediate grade at separate sites) at initial diagnosis [7]. In contrast, patients with either an antecedant low grade lymphoma or known discordant histology at primary diagnosis were not included in our study cohort. In the Dana-Farber report two of their eleven late relapses (>24 months post treatment) were low-grade. The 11% of relapses with low-grade histology noted in our series may represent the de novo occurrence of a second lymphoma or the low grade histology may have been present at diagnosis and
missed. Consistent with this latter hypothesis is that 18% of patients who had multiple biopsies (sites and number of sites were not specified) at the National Cancer Institute between 1953 and 1975 were recognized as having discordant histologies [22]. Late relapse of diffuse large cell lymphoma (more than two years from diagnosis) is believed by most oncologists to be an an unusual occurrence. This is an erroneous belief. Twenty-five per cent of the intermediate grade relapses occurred more than two years after the initial diagnosis with the latest being in the eleventh year. This is similar to the Dana-Farber Experience with M-BACOD in which 32% of the relapses, including low grade, occurred after two years [8]. Overall our late relapse rate was 7% (29 of 402) which is similar to that in other series [7, 8]. This will be an under estimate of the true rate of late relapses, given that they occured up to 11 years from diagnosis, as our median follow up was only 4.25 years (range 0.3-16 years). Late relapses may be due to de novo lymphoma arising in a susceptible host with pre-existing chromosomal abnormalities or it may represent recurrence of the original lymphoma. Three patients in the M.D. Anderson series (relapsing at 2.5-4 years) had tissue from diagnosis and relapse available for comparison. Phenotype and DNA content were the same implying recurrence [7]. In an earlier report of one case from our group Southern blot analysis of immunoglobulin heavy chain rearrangement demonstrated an identical pattern for the relapse (at 57 months) and the original tumour [23], Contrary to common belief, death from lymphoma, after relapse, is not inevitable. Overall survival after relapse was a median of 1.2 years with afiveyear rate of 24%. For failure free survival the corresponding figures were 0.6 years and 18%. For those patients with intermediate grade relapse the median overall survival was one year with a 20% five year rate; for failure free survival 0.6 years and 17%. Long term survival is therefore possible and appropriate therapy for relapsed inter-
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
Symptoms0 LDHC Number of nodal sites0 Number of extranodal sites0 Number of extranodaJ sites'1
Adverse factor
1130 a)
100=1
p = <0 001
p=00O2
20-
S ^ - - ^ * . . , . . . ,
4
6
10
b)
100 ?
80 p = 0001
C) 1
10
0
2
4
6
8
10
r
Figure 4. Overall survival for intermediate grade relapses from relapse. a) LDH value at relapse $ upper limit normal (n = 32) > upper limit normal (n = 32) b) Number of extranodal sites at primary diagnosis None (TI = 44) One(7i = 23) More than one (TI = 16) c) Symptoms at relapse A symptoms (TI = 65) B symptoms (TI = 18)
first
mediate grade disease should be instituted. High dose chemotherapy with marrow transplantation has been shown to be the treatment of choice in selected younger patients with chemosensitive disease (11). The results of the Parma trial with, after response to initial DHAP therapy, a randomization between either high dose chemotherapy/autologous bone marrow transplantation or further chemotherapy with DHAP have given further
Figure 5. Overall survival from first relapse. ) Performance status (ECOG) 0 (n = 30) 1 (n = 32) 2 (n = 14) 3 (n = 6) b) Time from initial diagnosis to first relapse < 1 year (n = 44) > 1 year (TI = 40) c) Number of nodal sites at relapse < 3 (n = 62) > 3 (TI = 20)
a
support to this approach [24]. However, a caveat is needed as evidenced by the outcomes reported in this paper. The relapse therapies chosen for the patients in this report were selected by their physicians based upon their perception of the patients likely outcome. Six of the twelve who had apparently localized relapse received combination chemotherapy and involved field irradiation and are still alive in remission. Similarly three of
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
Year
1131
In conclusion, 24% of non-elderly patients treated with 12-week combination chemotherapy for advanced diffuse large-cell lymphoma who entered a complete response relapsed with the majority of relapses within two years of original diagnosis. However, 25% of the relapses were delayed with the latest recorded in the eleventh year. Eleven percent of these first relapses were of low-grade histology which has therapeutic implications and demonstrates the value of biopsy verification. Independent prognostic variables were identified at relapse predictive of outcome. These need to be validated and then, if reproducible, accounted for when analyzing results of salvage therapy.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 15.
16.
Acknowledgements
Jan Wilson for her secretarial help and the physicians of the Province of British Columbia who referred the patients.
References 1. Weick JK, Dahlberg S, Fisher RI et al. Combination chemotherapy of intermediate-grade and high-grade non-Hodgkin's lymphoma with MACOP-B: A Southwest Oncology Group Study. J Clin Oncol 1991; 9: 748-53. 2. Longo DT, Devita VT, Duffey PL et al. Superiority of ProMACECytaBOM over ProMACE-MOPP in the treatment of advanced diffuse aggressive lymphoma: Results of a prospective randomized trial. J Clin Oncol 1991; 9: 25-38. 3. Dana BW, Dahlberg S, Miller TP et al. m-BACOD treatment for intermediate and high grade malignant lymphomas; A South-
17.
18.
19.
20.
21.
22.
west Oncology Group phase II trial. J Clin Oncol 1990; 8: 1155— 62. Fisher RI, Miller TP, Dana BW et al. Southwest Oncology Group clinical trials for intermediate and high grade non-Hodgkin's lymphoma. Semin Haematol 1987; 24: 21-5. Fisher RI, Gaynor ER, Dahlberg W et al. Comparison of a standard regimen (CHOP) with three intensive chemotherapy regimens for advanced non-Hodgkin's lymphoma. N Engl J Med 1993; 328: 1002-6. Coiffier B, Gisselbrecht C, Herbrecht R et al. LNH-84 Regimen: A multicenter study of intensive chemotherapy in 737 patients with aggressive malignant lymphoma. J Clin Oncol 19889; 7: 1018-26. Cabanillas F, Velasquez WS, Hagenmeister FB et al. Clinical, biological and histologic features of late relapses in diffuse large cell lymphoma. Blood 1992; 79: 1024-8. Shipp MA, Yeap BY, Harrington DP et al. The m-BACOD combination chemotherapy regimen in large-cell lymphoma. Analysis of the complete trial and comparison with the M-BACOD regimen. J Clin Oncol 1990; 8: 84-93. Coltman CA, Dahlberg S, Iches SE et al. CHOP is curative in thirty percent of patients with large cell lymphoma. A twelve year Southwest Oncology Group follow-up. In Skarin AT (ed): Update on Treatment for Diffuse Large Cell Lymphoma. New York: Park Row 1986; 71-81. Philip T, Hagenbeek A, Gomez F et al. Salvage treatment for chemotherapy resistant diffuse non-Hodgkin's lymphoma (NHL). Review of the 215 patients of the Parma study and rationale for a second prospective study. Ann Oncol 1986; 7 (Suppl 3): 17 (Abstr 54). Bosly A, Coiffier B, Gisselbrecht C et al. Bone marrow transplantation prolongs survival after relapse in aggressive lymphoma patients treated with the LNH-84 regimen. J Clin Oncol 1992; 10: 1615-23. Blay J-Y, Gomez F, Bachelot Tet al. The international prognostic index correlates to survival in intermediate/high grade lymphoma (NHL) in relapse: Analysis of the Parma trial. Proc Am Soc Clin Oncol 1996; 15: 151 (Abstr 269). Hoskins P, Connors JM, Gascoyne R et al. 10-year outcome of 12-week duration chemotherapy for advanced large-cell lymphoma: The Vancouver experience. Ann Oncol 1996; 7: 58 (Abstr 198). Kalbfleisch JD, PTentice RL. The Statistical Analysis of Failure Time Data. New York: Wiley 1980. Non-Hodgkin's Lymphoma Pathologic Classification Project: National Cancer Institute sponsored study of classifications of non-Hodgkin's lymphoma. Summary and description of a working formulation for clinical usage. Cancer 1982; 49: 2112-35. Marazuela M, Yebra M, Giron JA et al. Late relapse with nodular lymphoma after treatment for diffuse non-Hodgkin's lymphoma. Cancer 1991; 67: 1950-3. Siegel RD, Skarin AT, Canellos GP. Relapse with nodular lymphoma following combination chemotherapy for diffuse nonHodgkin's lymphoma. Am J Clin Oncol 1986; 9: 420-3. O'Brien CJ, Howard M, Child JA. Downgrading of non-Hodgkin's lymphoma following chemotherapy. Acta Haematol 1987; 77: 57-9. Head DR, Avakian J, Kjeklsberg R, Cerezo L. Relapse of intermediate or high-grade (unfavorable) non-Hodgkin's lymphoma as a low-grade (favorable) non-Hodgkin's lymphoma. Am J Clin Pathol 1988; 289: 106-8. Cheragan DP, Four K, Dressier L. High grade non-Hodgkin's lymphoma relapsing as low-grade follicular lymphoma. So-called downgraded lymphoma. Am J Hematol 1989; 30: 36-4. Mead GM, Mackintosh FR, Burke JS, Rosenberg SA. Late relapse from complete remission in nodular and diffuse histiocytic lymphoma. Cancer 1983; 52: 1356-9. Fisher RI, Jones RB, DeVita VT et al. Natural history of malignant lymphomas with divergent histologies at staging evaluation. Cancer 1981; 47: 2022-5.
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019
eight treated with irradiation alone are in continuing remission. Patient selection alone, regardless of the treatment used, may alter outcome. It is therefore essential that in randomized trials the patient arms be equivalent with regard to their prognostic mix. Therefore, in an attempt to identify which factors identifiable at relapse most influenced outcome we performed univariate and multivariate analysis in the patients with intermediate grade relapse. Adverse independent factors both for overall survival and failure free survival were: a duration from initial diagnosis to first relapse of less than one year, decreasing performance status at relapse, and more than three nodal sites at relapse. These same factors were important whether or not the LDH value was included in the analysis. LDH was not an independent factor but it was only measured in 74% and so a true effect may have been missed. Limitations of this analysis were that uniform treatment on relapse was not used, that the extent of re-staging varied, and that elderly patients were not part of the initial treatment cohort. As such these results should only be regarded as hypothesis generating. The prognostic factor analysis does need to be repeated in data sets where the above limitations do not exist. Despite this the results are consistent with those from other such analyses performed at initial diagnosis [25] and at relapse [7,10-12].
1132 23. Lee AYY, Connors JM, Klimo P et al. Late relapse in patients with diffuse large cell lymphoma treated with MACOP-B. J Clin Oncol 1997; 15: 1445-1753. 24. Philip T, Guglielmi C, Hagenbeelc A et al. Autologous bone marrow transplantation as compared with salvage chemotherapy in relapses of chemotherapy-sensitive non-Hodgkin's lymphoma. N Engl J Med 1995; 333: 1540-5. 25. The International Non-Hodgltin's lymphoma Prognostic Factors Project. A predictive model for aggressive non-Hodgkin's lymphoma. N Engl J Med 1993; 329: 987-94.
Received 20 June 1997; accepted 25 August 1997.
Correspondence to: Dr. P. J. Hoskins British Columbia Cancer Agency Div. of Medical Oncology 600 West 10th Avenue Vancouver British Columbia, Canada V5Z4E6
Downloaded from https://academic.oup.com/annonc/article-abstract/8/11/1125/146411 by guest on 23 January 2019