Prognosis with newly diagnosed mycosis fungoides after total skin electron radiation of 30 or 35 GY

Int. J. Radiation

Pergamon

Oncology Biol. Phys., Vol. 28, No. 4, pp. 839-845, 1994 Copyright 0 1994 Elsevier Science Ltd Printed in the USA. All rights reserved 0360-30 I6/94 $6.00 + .OO

??Clinical Original Contribution

PROGNOSIS WITH NEWLY DIAGNOSED MYCOSIS FUNGOIDES AFTER TOTAL SKIN ELECTRON RADIATION OF 30 OR 35 GY GLENN W. JONES, FRCPC,1,2 ALFIE TADROS, M.B.,’ DAVID I. HODSON, FRCPC,1>2 DONALD ROSENTHAL, FRCPC,3 JOHN ROBERTS, FRCPC4 AND BARBARA THORSON, RN1 ‘Hamilton Regional CancerCentre,Ontario CancerFoundation; ‘Departmentof Medicine, 3Departmentof Dermatology, McMasterUniversity,4Departmentof Pathology, Hamilton Civics Hospitals,Hamilton, Ontario,Canada; ‘Nova Scotia CancerCentre Purpose:To determinethe prognosis of new patients with TI-~No_IB&& mycosis fungoides treated with total skin electron beam radiation. Methods and Materials: 25 consecutive patients received 30 Gy with 3 or 4 MeV electrons in 1977-1980; 121 received 35 Gy with 4 MeV in 1980-1992. Response rates, relapse-free survival, and overall and cause-specific survivals were assessed by explicit criteria. The relationships of T, N, gender, age, and radiation technique to prognosis were investigated by regression statistics. Results: The average age was 55 years and the male:female ratio was 1:4. Forty-four percent were TIN0 and 34% were T*No. The overall complete response rate was 8246, and lower T status, more radiation, and female gender were independently and positively associated with response. Median follow-up was 5.2 years. T1 patients who entered remission had a higher relapse-free survival compared to T2 through T4 patients. Thirty-four percent of T, patients remained relapse-free at 6 years, compared to fewer than 20% of TEA patients. For all 146 patients the median overall survival was not reached at 15 years. Only 8 of 29 deaths were related to mycosis fungoides and these were significantly associated with higher T. The 54 TIN0 patients who had 35 Gy had a IO-year mycosis fungoides-specific survival of 100%. Conclusion: Total skin electron beam radiation gives good results with TIN&M,, disease. TM disease is less likely to respond, it relapses more quickly, and it implies a poorer survival, but radiation offers palliation. T2 responds like TI , but relapses like TM. T2 also implies an intermediate survival. These results have implications for staging, informed consent, optimizing radiation treatment, and clinical trials. Mycosis fungoides, Radiation, Staging, Prognosis, Survival.

INTRODUCIION

and the issue of which treatment should be first remains controversial (3, 11, 13, 14). Total skin electron beam (TSEB) radiation treatment was first reported in 1953 (19). Subsequently, a number of centers in Europe, America, and Canada began TSEB treatments. The best known American series is from Stanford (4,7). In Canada, the first patient was treated at Hamilton in 1969 ( 17). Hamilton pursued a systematic escalation in dose and energy. Thirty Gy with 3 MeV electrons began in 1973; 35 Gy with 4 MeV, in 1980. In 1984, the Canadian Dermatology Association and Hamilton formed the Canadian Co-operative Mycosis Fungoides Study Group to promote clinical research in mycosis fungoides and to continue a TSEB-based management strategy for the majority of Canadian patients. By then, over 150 patients from across Canada had received TSEB in Hamilton and the results from the first 106 had been published (17). Since 1969, our policy has been to

Mycosis fungoides is an uncommon cutaneous T-cell lymphoma ( 12, 18). Believed to start in the epidermis, it usually follows an indolent clinical course over many years, exhibiting a progressive increase in the numbers of skin lesions and in the degree of skin infiltration. Some patients may develop tumorous skin nodules (7, 12). Eventually, internal spread to visceral organs may occur by lymphatic or hematogenous routes. Patients are usually given a diagnosis when disease is clinically limited to the skin. Initial treatment is directed to the skin with curative intent, and usually consists of either topical chemotherapy, such as mechlorethamine or photoactivated psoralen, or ionizing radiation with low energy electrons (total skin electron beam radiation). Properly designed randomized trials comparing these modalities as first-line therapy for early stage mycosis fungoides have not been performed

Reprint requests to: Glenn W. Jones, 699 Concession St., Hamilton, Ontario, L8V X2 Canada.

Acceptedfor publication 1 October 1993. 839

840

I. J. Radiation Oncology 0 Biology 0 Physics

offer TSEB to all patients on their referral, if they are technically eligible for TSEB. We have previously demonstrated that patients who receive TSEB after failing a previous therapy, such as mechlorethamine, psoralen and ultraviolet A, systemic chemotherapy, and locoregional radiation, have a worse response and prognosis as compared to newly diagnosed patients (9). In this report we estimate the rate of complete remission, the relapse-free survival, and the overall and cause-specific survivals for newly diagnosed cases treated with TSEB alone. The results of two radiation techniques are compared and T, N, Stage, age, and gender are investigated for possible associations with prognosis.

METHODS

AND MATERIALS

Since 1980, patients have been assessed in an interdisciplinary clinic of dermatologists, radiation oncologists, medical oncologists and a pathologist. We include interview and physical examination, review by two dermatopathologists of all previous pathology and new biopsies of skin, complete blood work, and chest radiography. After 1977, all patients had skin specimens sent for electron microscopy and peripheral blood sent for Buffy coat to identify Sezary cells. Suspicious lymph nodes are biopsied. Until 1990 most patients underwent some combination of bipedal lymphangiography, computed tomography of the abdomen and pelvis, liver-spleen scintigraphy, and bone marrow biopsy. Most patients had all four investigations. An audit demonstrated no positive result in newly diagnosed cases with early skin involvement and for these patients the tests were discontinued. We assign Stage at referral with a version of the unofficial TNM-like classification system (1, 12). T denotes the type and extent of skin involvement and N and M denote nodal and visceral involvement, respectively. Briefly, T, = patch-plaque disease less than 10% of skin surface area, T2 = patch-plaque 10% or more of skin surface area, T3 = one or more tumors on the skin, and T4 = generalized erythroderma; No = no palpable nodes, N, = palpable nodes but benign, Nz = palpable with dermatopathic lymphadenopathy on biopsy, and N3 = palpable with pathologic involvement; M1 = biopsy confirmed visceral disease. B,J indicates that blood exhibits no Sezary cells by Buffy coat or on peripheral blood differential, and B, indicates any circulating Sezary cells. Stage groups are defined as IA (TINoMO), IB (T2NoM0), IIA (TI-zNIILIO),IIB (T~No-Mo), 111 (&NO-MO), WA (any N2-jMo), and IVB (any Mi). In these groups, B (as opposed to A) neither signifies B, status nor lymphomatous Bsymptoms; neither are relevant for staging. At consultation we additionally classify a patient according to whether he or she has failed an oncologic therapy prior to TSEB, or whether he or she receives TSEB at diagnosis. All technically eligible patients were offered TSEB after the diagnosis of mycosis fungoides was confirmed. Total

Volume 28, Number 4, 1994

skin electron beam radiation was delivered as previously described (9, 17) and since 1980 it has been unchanged. Since 1980, we have delivered 35 Gy in 12 fractions in 3 weeks to the whole skin with 4.0 MeV electrons. The energy is defined in accordance with the ICRU (8). We have used a composite, six field arrangement of one anteroposterior and two oblique pairs, treating one pair per day on sequential days, 4 days-a-week. The half of the patient’s surface which faces the machine has been treated in one horizontal and stationary field, centered on the patient’s midsection. Internal eye shields of machined steel, with clay goggles over the eyelids, were used throughout treatment in all patients. Hands were shielded with radiographic gloves and ankles and feet were shielded with lead strips during two or three antero-posterior pair treatments. Lips and groin were shielded using lead strips of appropriate shape in the last two or three fractions of treatment. There were no other areas of shielding and no areas were boosted with additional radiation beyond TSEB. Complete clinical response to TSEB was assessed 3 months after the first fraction of radiation. An assessment at 3 months avoids the acute radiation effect and classifies patients as nonresponders when they experience a very transient remission. In clinically negative patients, skin biopsies were taken from sites of previous clinical and pathologic disease. If positive, these were repeated every 3 months until negative or until clinical relapse. A positive biopsy in the absence of clinical lesions was not called a treatment failure and did not lead to additional treatment. It is our experience that positive biopsies at 3 months are followed by negative biopsies at 6 months in patients who do not have overt clinical lesions at, and between, those times. A nonresponse was any persistent clinical disease, regardless of amount, as seen at 3 months and as confirmed on biopsy. Patients were formally followed on a quarterly to semiannual basis and were also seen on request. Relapse was defined as a return of any amount of clinical disease as confirmed by biopsy (routinely performed within a month of new clinical lesions). The first treatment initiated after nonresponse or relapse was nitrogen mustard, in most cases. A minority received limited-field and superficial radiation, psoralen with ultraviolet A, topical or systemic steroids, a retinoid, or a retinoid and psoralen with ultraviolet A. This depended upon the geographic availability of a therapy, in relation to the patient’s home, and the patient’s own preference. We attempted to contact all lost patients directly, or through family and referring physicians. Names of those who remained lost were linked to the Ontario death registry. Causes of death were classified as relating to mycosis fungoides (chemotherapy death, sepsis from skin involvement, or visceral complication), other malignancy (biopsy confirmed exclusion of progressive mycosis fungoides), cardiovascular events (myocardial infarction, stroke or embolus), and other well described causes. In most cases autopsies were unavailable, but the clinical courses were well documented.

Mycosis fungoides and TSEB 0 G. W.

Data on each patient before 1988 were extracted by retrospective chart review. Since 1988, all data have been collected with same day entry and verification on a data-base system.’ Bio-Mathematical Data Processing programs’ were used for all statistical analyses. Univariate associations with response were assessed by cross-tabulation and Chi-square. Event-free survival times were measured from the first day of TSEB. Corresponding plots were derived by the Kaplan-Meier method and two and multi-way comparisons used Mantel-Cox and generalized Savage product-moment methods. Multivariate modelling included logistic regression for response and Cox methods for event-free survivals. Logistic goodness-of-fit was determined by the Hosmer-Lemeshow Chi-square method (where a high p-value indicates a good fit) and by plotting influence and change in Chi-square diagnostics by proportion for each covariate pattern. Ninety-five percent confidence intervals on median event-free times were given by the Brookmeyer-Crowley method. All p-values are two-sided and the term statistical significance means p < 0.05. RESULTS Selection and characteristics of 146 study cases

Between 1955 and 1992, 442 patients have been referred to Hamilton with a confirmed diagnosis of mycosis fungoides. One hundred forty-six are the subject of this report. Reasons for excluding the remaining patients include the following: 109 were not offered TSEB or declined to have it, four were still waiting to receive TSEB as of March 1992, 16 had adjuvant mechlorethamine or systemic chemotherapy in 1980,26 had concurrent and adjuvant etretinate in 1987-1989 (9), two patients had no follow-up beyond the last fraction of TSEB, 12 1 received TSEB as a second or subsequent therapy, ten newly diagnosed patients had advanced disease (Nz-~ or B1 or M, cases), five newly diagnosed patients were treated strictly for palliation (three with comorbid medical, and two with severe psychiatric conditions), and three newly diagnosed patients received neither 30 nor 35 Gy for unspecified reasons (one each in 1972, 1974, and 1976). The 146 consecutive and newly diagnosed cases in this report received TSEB between 1977 and early 1992. One of these patients, with known heart disease, was followed for less than 3 months. He expired from congestive heart failure at 5 weeks and had no evidence of response. He was classified as a T2No nonresponder. The other 145 patients had a minimum follow-up of 3 months. All 146 cases were TI_4NO-IB~M0and 25 patients received 30 Gy (group A) and 121 patients received 35 Gy (group B). The electron energy for 24 group A patients was 3 MeV, and for the other 122 patients it was 4 MeV.

‘MedLog, Information Analysis Corporation, 480 San Antonio Rd., Suite 230, Mountain

View, CA.

JONES et al.

841

Table 1. Numbers of patients by T and N combinations and stages for both groups A and B N Group A 30 Gy n = 25 B

T

No

N,

T, T2

10 6

7-3

4

T4

1

2 2 0 0 2 7 5 3

T1 35 Gy n= 121

T2 T3 T4

54 44 4 2

Stages 10’A 6 IB

54’A 44’B

#IA

4m lH’ 9llA

9 IIB 5 111

The average age of all patients was 55 years (median = 56, range 23-85, 50% between 45 and 67). The age distributions of groups A and B were virtually identical. The male to female ratio was 1:4 (86:60) but the ratio was higher in group A (2.6 = 18:7). Table 1 separately shows the T and N combinations and Stages for group A and group B. For all 146 cases, the distribution in T was 47% T, , 40% TZ, 9% T3, and 4% T4. The distribution of Stage was 44% IA, 34% IB, 9% IIA, 9% IIB, and 4% III. There was no significant difference between groups A and B. Response rates

The combined complete remission rate was 81.5%. It was 64% ( 16/25) for group A and 85% ( 103/ 12 1) for group B. This difference was statistically significant at p = 0.0 13. Males were less likely to respond than females: 65/86 = 76% males compared to 54/60 = 90% females; p = 0.03. Response according to T was as follows: T1 60/68 = 88%, T2 47159 = 80%, T3 8113 = 62%, and T4 416; p = 0.09 for trend. Response was not significantly associated with N: No 104/129 = 83%, N, 15/21 = 71%; p = 0.2. The potential effect of T was mirrored in Stage: IA 57/64 = 89%, IB 41/50 = 82%, IIA 9/l 3 = 69%, IIB 8/13 = 62%, and III 416; p for trend = 0.09. Logistic regression revealed no statistical interactions. Response was linear in each variable. Univariate p-values were 0.02 10 for T, 0.2 for N, 0.023 for female gender, 0.3 for age in years, and 0.0206 for radiation technique. The multivariate adjusted model included T with p = 0.023, Gy with p = 0.042, and female gender with p = 0.041. Regression diagnostics showed this to be a well behaved model. The goodness-of-fit for this model was excellent with p = 0.84 on 6” of freedom. T was the most associated independent variable with response. Comparing T4 to T, the odds ratio for response was 0.17, whereas the odds ratio for radiation (more vs. less) was 2.9 and that for gender (female vs. male) was 2.7. Odds ratios less than 1.O indicate a poorer response with increasingly positive

‘BMDP

Statistical

Software

Inc., Los Angeles, CA.

842

I. J. Radiation Oncology 0 Biology 0 Physics

Volume 28, Number 4, 1994

values of a variable, whereas ratios over 1.0 indicate a better response with increasingly positive values. This model predicts a 97% chance that a female with T, given 4 MeV electrons to 35 Gy will respond completely and predicts a 4 1% chance for a male with T4 who receives 3 MeV electrons to 30 Gy. General follow-up information

The average follow-up was 5.4 years (median 5.2). Because of the strict dose escalation in 1980 which defined groups A and B, group A had a longer follow-up (median 8.5, average 9.3, maximum of 17) as compared to group B (median and average both 4.6, maximum of 11). Two of 25 group A patients and seven of 12 1 group B patients were completely lost to follow-up (total = 9/id6or 6%). The relapse-free, overall, and cause-specific survival experiences of all 146 patients are shown in Figure 1. The median relapse-free survival for the 119 patients who entered remission was 2.8 years (95% confidence interval = 1.8 3.9). Despite this short duration, median overall survival in the entire 146 patients was estimated at over 15 years. Only eight of 29 deaths could be related to mycosis fungoides. Relapse-free survival

Seventy-one of the 119 who entered remission have relapsed (’‘/,6 in A, ““/iO,in B). Relapse-free survival plots for T, and T2, for which there are sufficient numbers, are shown in Figure 2. T,No_i patients who entered remission had an estimated 40% chance of being relapse-free at 5 years, with a 95% confidence interval (95CI) of 2.2-5.0 years on the median of 4.1 years. T2No_i relapse-free survival was 18% at 5 years with a median of 1.8 years (95CI = 1.3-2.8). A direct comparison by Mantel-Cox method between T, and Tz was significant with p = 0.02. In contrast, the results for Tz and T3_4were similar to each other. This pattern was explored by Cox models and relapsefree survival was associated differently with Ti and all TZ_4(p = 0.024). Thus, T was linearly associated with complete remission but it was not linearly associated with time to relapse. N, Gy, age, and gender were not associated with relapse-free survival (each p > 0.3). Overall survival

At last contact 20% of patients had expired, 14/25 in A and 15/ 12 1 in B. Group B had shorter follow-up, but had a similar age profile at the time of TSEB. The distribution of causes of death were similar in both groups. Four deaths in A and four in B were related to mycosis fungoides. Three group A patients died for “other” reasons which were one accident, one corpulmonale from obstructive lung disease and one bronchopneumonia. Three A and seven B patients died from cardiovascular events. Other malignancies claimed four A and three B patients: one nonHodgkins lymphoma in a patient in remission after TSEB, one larynx cancer in a patient in remission after TSEB, two rectal carcinomas and one nonsmall-cell

1

100

(fi)

(15)

90

1

MFRDFS

80 s

70

5 =

60

Cn Y

50

E $

40

$ w

30

-I 0

5

10

15

YEARS FROM STARTING BEAM Fig. 1. Event-free survival. MFRDFS = Mycosis fungoides-related death-free survival (n = 146); OS = Overall survival (death from any cause, n = 146); RFS = Relapse-free survival (n = 119 who entered complete clinical remission). Numbers in brackets show patients remaining at risk at 2.5, 5, 10, and 15 years from start of TSEB.

lung cancer in three patients who had entered remission with TSEB, but who had relapsed with mycosis fungoides,

one small-cell lung cancer in a patient who did not respond to TSEB, and one leukemia. The latter patient had received chemotherapy for a nonHodgkins lymphoma prior to the diagnosis of mycosis fungoides. He entered remission for mycosis fungoides with TSEB, but relapsed with mycosis fungoides. He subsequently developed acute myelogenous leukemia. For all 146 patients the 5 and IO-year survivals from deaths from any cause were 93 and 86% for T, , 8 1 and 69% for TZ, and 53 and 32% for Ts. Estimates are not meaningful for the six T4 cases, with two dead before 5 years and two alive between 5 and 8 years. The 5 and loyear survivals from deaths from any cause were 92 and 88% for Stage IA, 88 and 72% for IB, 64 and 55% for IIA, 53 and 32% for IIB. For the 29 deaths from any cause the univariate Cox modelling revealed death to be associated with age with p = 0.0006, T with p = 0.001, Gy with p = 0.0 18, and male gender with p = 0.037. Multivariate adjustment revealed only T (p = 0.014) and age (p = 0.011) to be significant. Cause-specific survival

Twenty-one of 29 deaths were not related to mycosis fungoides. Multivariate Cox regression for that outcome showed age to be the only significant factor (p = 0.001).

Mycosis fungoides and TSEB 0 G. W.

There were eight mycosis fungoides related deaths. T was most significantly associated with this event (p = 0.005). Figure 3 gives the cause (mycosis fungoides)-specific survival plots according to T. Of note is that only 1 of 68 T, patients died from mycosis fungoides and he was N1 and received 30 Gy. Thus the 54 Stage IA or TINoBoMo patients in group B experienced a mycosis fungoides-specific IO-year survival of 100%.

JONES

et

al.

843

(35)

100

I

(8)

T’

(32) ’

90 3

80-

5 5 cn

70-

ii !

60-

F

50-

i 0 P

40-

Y 4

30-

%

20-

DISCUSSION

For this report we have estimated prognosis following total skin electron beam radiation in 146 consecutive patients with a new diagnosis of T,-,+NO-,BOMO mycosis fungoides. Valid and precise estimates of prognosis are important for informed consent, for anticipating management difficulties, and for designing and interpreting clinical trials. The characteristics and prognoses of newly diagnosed skin-limited disease, newly diagnosed advanced disease, failed prior therapy and skin-limited disease, and failed prior therapy and advanced disease patient groups differ and should be reported separately (10). Here we report results of the first group, since this is the most important clinical group in which to judge the efficacy of radiation. To strengthen validity we have used explicit criteria for baseline characteristics, standard therapies and treatment outcomes, and we have minimized any loss to follow-up.

-l

T2

(2) ”

I

0

i

lb

1

YEARS FROM STARTING BEAM Fig. 2. Relapse-free survivals for T, and T2 disease (n = 60 and 47 who entered complete clinical remission, respectively). Numbers in brackets show patients remaining at risk at 2.5, 5, and 10 years from start of TSEB.

10 -

0

5

10

15

YEARS FROM STARTING BEAM Fig. 3. Mycosis fungoides-related death-free survivals according to T classification: 68 T, , 59 T2, 13 T3, and 6 T4 patients. Numbers in brackets show patients remaining at risk at 5 and 10 years from start of TSEB.

This study suggests that the chance of complete remission with TSEB increases with lower T, female gender and more intense radiation technique. As noted, a T, female receiving 35 Gy of 4 MeV electrons has an estimated chance of complete clinical remission of 97%. A comparison of our two sequential radiation techniques is subject to potential biases that can affect any historic cohort comparison (2). However, we found a significant gain of about 5% in response rates across all Stage and gender covariate combinations in favor of more aggressive radiation. The implication is that we might further optimize TSEB through additional changes in technique, electron energy, total dose and fraction size. Our study shows no apparent effect on relapse-free survival and causes of death of our more aggressive radiation technique as compared to our less aggressive technique. This may be because TSEB has no effect on the natural history of the disease. Alternatively, this may be a statistical limitation related to the numbers of cases in each subgroup, the small numbers of deaths, and the similarity of the two techniques. In particular, our study suggests that fewer than 50% of T,No_r patients may be cured and there is no evidence that any other Stages can be cured with radiation. While it is difficult to define cure-relapse can occur 8 or more years after remission from TSEBthe rate at which relapse is occurring in Stage IA is estimated to be declining each year after TSEB. That trend

844

I. J. Radiation Oncology 0 Biology 0 Physics

is not apparent in other Stages. Given a mycosis fungoidesspecific survival in Stage IA of 100% at 10 years, our overall statistical power was limited. We were unable to show any overall impact of a more aggressive radiation technique on survival in all 146 patients, in whom the median follow-up was but 5.2 years, and of whom only 44% were Stage IA. The relapse-free survivals of T2 through T4 patients were not statistically different from each other, although they individually and collectively differed from the experience of T, patients. Though T3_4 patients were few in number, there was no evidence that their experience was worse than that of T2 patients. This apparent nonlinear association of T with disease-free survival may have arisen by chance. Conversely, this might reflect the combined qualitative and quantitative aspects of the T descriptors. For example, patients with T3_4 who enter remission could have qualitatively and quantitatively similar disease to that in patients with Tz. The overall effect of radiation could be similar for these two groups of patients, despite one group’s tendency to form tumors or cause diffuse erythroderma. In contrast, T3-4 patients who do not enter remission after TSEB could have quantitatively or qualitatively different disease that expresses it’s different nature as nonresponse and poor survival after TSEB. Another explanation is that differences in response and prognosis might be due to disease more disseminated than is clinically apparent or detectable by conventional tests during staging. Such disease, perhaps more likely in Stages IB and higher, could recolonize the skin during and after TSEB and could result in immediate treatment failure or early recurrence. Finally, the T descriptors were initially defined only in relation to overall survival (1) and a linear relationship with overall survival need not imply a similar relationship with intervening events. Given an apparently complex relationship between T and several different outcomes, other baseline characteristics should be investigated for any associations with outcomes of cure, no response, and short times to relapse. They might provide additional independent information apart from T. A stratification within each T group could refine entry criteria and sample size specification for clinical trials and could affect management. There are a number of published reports that detail the prognoses of collections or series of patients receiving topical therapies (4, 15, 16, 20). Total skin electron beam radiation results from Stanford (4) and mechlorethamine results from other centers (15, 20) are of most interest. Direct comparisons with our results are not strictly appropriate. For those centers reporting results of TSEB, for example, there are major differences between centers in patient, tumor, and previous therapy characteristics, and in radiation dose and technique. We provide data exclusively on new patients with mycosis fungoides limited to skin, who received 35 Gy of TSEB, and who were assessed using explicit outcome criteria. Data from different centers will have to be published with similar subgroupings and with similar criteria to enable a direct comparison to be

Volume 28, Number 4, 1994

made. Nevertheless, we and Stanford have shown a consistent potential for cure in Stage IA, and we both report that doses higher than 30 Gy produce better results (6). While a comparison of Stanford and Hamilton TSEB data with two good mechlorethamine series (15, 20) would suggest that TSEB offers higher rates of remission, longer relapse-free survivals, and greater potential for cure, a potential superiority of TSEB awaits confirmation in a randomized trial. There are only two published randomized trials comparing therapeutic management strategies (11, 14). The American Cooperative Group reported on only 12 patients who entered a brief three arm study and the results are not helpful (14). The trial by Kaye et al. (11) included 103 randomized cases who received either combination radiation and systemic chemotherapy, or a sequence of less aggressive topical therapies which could culminate, by escalation, in radiation or chemotherapy, depending on clinical course. Only 45 patients had disease limited to skin and only six were T,No. Patients were eligible if they had received a previous therapy for mycosis fungoides and this factor was used to stratify patients for randomization. There was no detailed, published description of radiation technique, but patients were planned to receive, if tolerated, up to 30 Gy at an unspecified energy. This trial had too low a power to detect differences in outcomes across subgroups. In addition, one can question how strong a test of the primary question this trial truly was, given a preponderance of advanced cases, some limitations on radiation technique, and a proportion of crossovers to (several) therapies. The question of a best treatment strategy remains open. It would be appropriate to conduct another randomized trial to compare management strategies when TSEB is shown to be optimized and risk groups are more precisely defined. It should target new patients with disease apparently limited to skin. Mycosis fungoides is an uncommon disorder. Three-quarters of new patients are Stages IA and IB, with chances of complete remission over 80% and with survivals measured in decades. Therefore, the traditional outcomes of response and survival pose severe challenges for the design, conduct, and statistical analyses of a trial. Other outcome measures like cost and preference must be developed to aid decision-making. We have decided to continue to offer TSEB to new patients. This strategy may cure some newly diagnosed IA patients and it certainly gives symptomatic relief for IB-Ill patients. Most deaths are not currently associated with mycosis fungoides or its treatments. We are using our summary estimates of prognosis from group B to elicit patient preference. Since we are unable to prioritize firstline treatments by evidence from randomized trials, we believe that new patients should be informed of all management alternatives. We are investigating alternative staging classifications to better encode risk. We are also developing a more optimized TSEB technique and additional outcome measures.

Mycosis fungoides and TSEB 0 G. W.

JONES et al.

845

REFERENCES 1. Bunn, P. A.; Lamberg, S. I. Report of the committee on staging and classification of cutaneous T-cell lymphoma. Cancer Treat. Rep. 63:725-728; 1979. 2. Feinstein, A. R. Clinical epidemiology: The architecture of clinical research. Toronto: WB Saunders; 1985:226-232. 3. Hamminga, B.; Noordijk, E.; Van Vloten, W. A. Treatment of mycosis fungoides: Total-skin electron-beam irradiation vs. topical mechlorethamine therapy. Arch. Dermatol. 118: 150-153; 1982. 4. Hoppe, R. T. Total skin electron beam therapy in the management of mycosis fungoides. In: Veath, J. M., Meyer, J. L., eds. The role of high energy electrons in the treatment of cancer. Frontiers in radiation therapy and oncology, Vol. 25. Basel: Karger; 1991:80-89. 5. Hoppe, R. T.; Cox, R. S.; Fuks, Z.; Price, N. M.; Bagshaw, M. A.; Farber, E. M. Electron-beam therapy for mycosis fungoides: The Stanford University experience. Cancer Treat. Rep. 63:691-70X$ 1979. 6. Hoppe, R. T.; Fuks, Z.; Bagshaw, M. A. Radiation therapy in the management ofcutanteous T-cell lymphoma. Cancer Treat. Rep. 63:625-632; 1979. 7. Hoppe, R. T.; Wood, G. S.; Abel, E. A. Mycosis fungoides and the Sezary syndrome: Pathology, staging, and treatment. Curr. Prob. Cancer. 14:297-361; 1990. 8. International Commission on Radiation Units and Measurements. ICRU report 35. Bethesda, MD ICRU; 1984: 61-63. 9. Jones, G.; McLean, J.; Rosenthal, D.; Roberts, J.; Sauder, D. N. Combined treatment with oral etretinate and electron beam therapy in patients with cutaneous T-cell lymphoma (mycosis fungoides and Sezary syndrome). J. Am. Acad. Dermatol. 26:960-967; 1992. 10. Jones, G.; Tadros, A.; Rosenthal, D.; Sauder, D.; Roberts, J. The Hamilton experience in mycosis fungoides using total skin electron beam (TSEB), 198 1- 1987. Clin. Invest. Med. 13:Bll6; 1990. 11. Kaye, F.; Bunn, P. A.; Steinberg, S. M.; Stocker, J. L.; Ihde, D. C.; Fischmann, A. B.; Glastein, E. J.; Schechter, G. P.; Phelps, R. M.; Foss, F. M.; Parlette, H. L.; Anderson, M. J.; Sausville, E. A. A randomized trial comparing combination electron-beam radiation and chemotherapy with

12.

13.

14.

15.

16.

17.

18.

19.

20.

topical therapy in the initial treatment of mycosis fungoides. N. Engl. J. Med. 321:1784-1790; 1989. Kuzel, T. M.; Roenigk, H. H.; Rosen, S. T. Mycosis fungoides and the Sezary syndrome: A review of pathogenesis, diagnosis, and therapy. J. Clin. Oncol. 9: 1298- 13 13; 199 1, Kuzel, T. M.; Springer, E.; Roenigk, H. H.; Rosen, S. T. Mycosis fungoides: Time to define the best therapy. JNCI 82:183-185; 1990. Lamberg, S. I.; Green, S. B.; Byar, D. P.; Block, J. B.; Clendenning, W. E.; Epstein, E. H.; Fuks, Z. Y.; Golitz, L. E.; Lorincz, A. L.; Michel, B.; Roenigk, H. H.; Van Scott, E. J.; Vonderheid, E. C.; Thomas, R. J. Status report of 376 mycosis fungoides patients at 4 years: Mycosis Fungoides Cooperative Group. Cancer Treat. Rep. 63:70 l-707; 1979. Ramsey, D. L.; Halperin, P. S.; Zeleniuck-Jacquotte, A. Topical mechlorethamine therapy for early stage mycosis fungoides. J. Am. Acad. Dermatol. 19:684-691; 1988. Reddy, S.; Parker, C. M.; Shidnia, H.; Gaffney, M. G.; Blyton, B. D.; Pruner, L. M.; Kalasinski, L. A. Total skin electron beam radiation therapy for mycosis fungoides. Am. J. Clin. Oncol. 15:119-124; 1992. Tadros, A. A. M.; Tepperman, B. S.; Hryniuk, W. M.; Peters, V. G.; Rosenthal, D.; Roberts, J. T.; Figueredo, A. T. Total skin electron irradiation for mycosis fungoides: Failure analysis and prognostic factors. Int. J. Radiat. Oncol. Biol. Phys. 9:1279-1287; 1983. The NonHodgkin’s Lymphoma Pathologic Classification Project: National Cancer Institute-sponsored study of classifications of nonHodgkin’s lymphomas: Summary and description of a working formulation for clinical usage. Cancer 49:2112-2135; 1982. Trump, J. G.; Wright, K. A.; Evans, W. W.; Anson, J. H.; Hare, H. S.; Fromer, J. L.; Jacque, G.; Home, K. W. High energy electrons for the treatment of extensive superficial malignant lesions. Am. J. Roentgenol. 69:623-629; 1953. Vonderheid, E. C.; Tan, E. T.; Kantor, A. F.; Shrager, L.; Micaily, B.; Van Scott, E. J. Longterm efficacy, curative potential, and carcinogenicity of topical mechlorethamine chemotherapy in cutaneous T cell lymphoma. J. Am. Acad. Dermatol. 20:4 16-428: 1989.