INT. J . HYPERTHERMIA,
1992,
VOL.
8,
NO. 3,
309-320
Response of canine soft tissue sarcomas to radiation or radiation plus hyperthermia: a randomized phase I1 study
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
S. McCHESNEY GILLETTEt II, M. W. DEWHIRSTg, E. L. GILLETTEt, D. E. THRALLI, R. L. PAGEI, B. E. POWERS?, S. J. WITHROW$, G. ROSNERS, C. WONG* and D. A. SIM* tDepartment of Radiology and Radiation Biology, and $Department of Clinical Sciences, Colorado State University, Fort Collins, Colorado 80523, USA $Division of Radiation Oncology, Duke University Medical Center, Durham, North Carolina 27710, USA fCollege of Veterinary Medicine, North Carolina State University, Raleigh, North Carolina 27606, USA *Clinical Trials Resources Group, John P. Robarts Research Institute, PO Box 5015, London Ontario, N6A 5KB Canada (Received 25 October 1991)
Sixty-four dogs with spontaneous soft tissue sarcomas without evidence of metastases were stratified by tumour volume and randomized to receive graded doses of radiotherapy (XRT) alone or radiotherapy plus hyperthermia (HT). An improvement in duration of local control was achieved with the addition of hyperthermia as compared with XRT alone (Wilcoxon, p=0.040; log rank, p=0.064). Overall frequency of late complications was not different for the two treatment arms when comparing across equivalent XRT dose groups. Frequency of distant metastases after therapy completion was not significantly different for the two treatment arms at 1 year (7.4% for XRT versus 20% for HT plus XRT) or 2 years ( I 1 . 5 % for XRT versus 25% for HT plus XRT) post therapy. These results suggest that a therapeutic gain was achieved for this group of turnour-bearing animals. Uni- and multivariate analyses were performed to examine the potential for various factors to influence treatment outcome. Patient related variables included tumour stage, histologic subtype and grade and tumour site. Treatment related variables included total radiation dose and 15 descriptors of temperature distributions achieved during hyperthermia. When considering parient related factors, tumour histology, grade and location were important predictors of time to minimum volume, but only turnour location influenced time to tumour regrowth. When considering treatmen/ related factors, radiation dose was not significantly correlated with time to minimum volume or time to local regrowth, but it was correlated with probability for late normal tissue damage in the XRT alone group (p=O.OOS). For the hyperthermia treatments, I3 of 15 tumour temperature distribution descriptors were correlated with time to minimum volume, but none were correlated with time to local regrowth. These results suggest that caution should be used in interpreting the value of temperature distribution descriptors in predicting for long-term local control after hyperthermia and radiotherapy. based on analysis of short-term responses. Key words: hyperthermia, radiation, canine sarcomas
1. Introduction Two Phase I1 studies using radiotherapy plus hyperthermia (HT) as a combined modality with graded doses of radiotherapy (XRT) or XRT plus surgery of human patients with soft tissue sarcomas have been published recently (Egawa et al. 1989, Leopold et al. 1989 and 1992). These studies evaluated the initial response to treatment as measured by changes IlCorrespondence to: Dr Sharon M. Gillette, Comparative Oncology Unit, Colorado State University, Fort Collins, Colorado 80523, USA 0265-6736/92 $3.00 01992 Taylor & Francis Lrd
S. McChesney Gillette et al.
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
310
in tumour volume and histologic parameters. In one study, four of 10 superficial tumours had a complete response to combined treatment (Egawa ef al. 1989). In a separate study, 34 patients with stage IKIB-IIA soft tissue sarcomas were randomized preoperatively to receive one or two fractions of HT per week, in conjunction with conventionally fractionated radiotherapy, to a total dose of 50 Gy (Leopold ef al. 1992). The complete response rate was relatively low in this patient population at 6 weeks after therapy completion. However, the group receiving two fractions of heat per week had a significantly greater percentage of patients with complete necrosis in surgically excised specimens than those patients receiving one fraction of heat per week. Furthermore, several descriptorsof the temperature distribution were positively correlated with percent necrosis (Oleson et al. 1989, Leopold et al. 1992). While these results suggest that HT can confer a therapeutic advantage in the treatment of soft tissue sarcomas, there continues to be a need for studies which address more relevent clinical endpoints, such as duration of local control, late normal damage and frequency of distant metastasis. Use of spontaneous tumours in pet animals has proven to be a good model system for testing the efficacy of new treatment modalities. Several studies using HT alone or combined with XRT have been done using spontaneous tumours (Marmor et al. 1978, Gillette et al. 1987, Dewhirst et al. 1983 and 1985, Brewer and Turrel 1982, Richardson et al. 1984). Three studies specifically evaluated canine soft tissue sarcomas (Dewhirst et al. 1983, Brewer and Turrel 1982, Richardson et al. 1984). All of the studies suggested an improvement in local control rates over that achievable with radiation alone, but small numbers of animals were accrued and thermometry was limited. In the present study, long-term follow-up was done on 64 spontaneous canine soft tissue sarcomas randomized to receive XRT alone or combined with HT with multipoint thermometry. This was a cooperative study between the Comparative Oncology Unit at Colorado State University (CSU) and the North Carolina Animal Cancer Treatment Program at North Carolina State University (NCSU). Hyperthermia treatments for NCSU dogs were done at Duke University Medical Center (DUMC). The emphasis of this analysis was to establish whether HT conferred any benefit over XRT alone, using short-term tumour regression, long-term local control, metastases and late normal tissue damage as endpoints. These endpoints were also evaluated for their correlation with patient and treatment related prognostic factors. 2. Materials and methods 2.1 . Experimental design Sixty-six dogs with soft tissue sarcomas were admitted to CSU or NCSU. All dogs were screened for metastatic disease by physical exam and chest radiographs. Results of a history, physical exam, CBC and blood chemistry panel were used to determine whether each animal was healthy enough to withstand treatment, which requires multiple anaesthesia procedures. Histologic type was confirmed by biopsy on all tumours before entry into the study. Two grading schemes were used to classify the tumours (Costa et al. 1984, Coindre et al. 1988). Criteria for exclusion included tumour volume >250 cm’, tumour depth > 4 0 cm, tumour margins larger than available hyperthermia applicator size, prior radiotherapy, chemotherapy or hyperthermia to the study lesion, and evidence of distance metastases. Tumours were also staged according to World Health Organization criteria (T, < 2 cm diameters, T, 2-5 cm diameter, T,>5 cm diameter or invasion of subcutis, T, tumour invading other structures) (WHO 1979). Two dogs were excluded from the final analysis. One patient was omitted because of thermal toxicity that occurred during the first hyperthermia treatment which necessitated amputation. A second patient was omitted because it developed metastases before the end of treatment. A third tumour was
-
Heat plus irradiation of sarcomas
311
Table 1. Dogs randomized into the study. Radiation Dose Radiation plus heat GY) only 5 6 5
47.5 50.0
3 7 0 7 4 5
52.5 55.0
5
2
0 0
33
31
35.0 40.0 42.5 45.0
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
Total
5 5 5
not considered because it was the second of two tumours occurring on the same animal. Dogs were stratified by tumour volume and randomized to receive total radiation doses between 35 and 55 Gy with or without hyperthermia (Table 1). Radiation was delivered in 10 fractions on a Monday, Wednesday and Friday schedule lasting 22 days. At CSU dogs were irradiated using a clinical linear accelerator which produced 6 MV X-rays. At NCSU a 6oCoteletherapy unit was used for irradiation. All irradiation and hyperthermia treatments were done under general anaesthesia using thiopental for induction and isoflurane for maintenance. Treatment planning using an Artronix treatment planning computer was done at both institutions. The prescribed tumour dose was defined as the minimum tumour dose. Maximum tumour dose never exceeded minimum dose by more than 15% . A 2-cm margin of normal tissue was included in the treatment volume. Hyperthermia was given approximately 3 h following irradiation. The target temperature was a minimum tumour temperature of 42°C without exceeding 40°C in normal tissues. At CSU hyperthermia was delivered with a 2 a3 MHz ultrasound transducer. Thermometry was done using a total of 10 thermistors in tumour and normal tissues (Gillette et al. 1987). Duration of HT was 30 min from the time that any intratumoral temperature reached 42°C. HT treatments were given twice weekly on a Monday-Friday schedule. At CSU the mean number of thermocouples was 4 . 6 & 1 - 4 in normal tissues and 5 - 4 + 1.4 in tumour. Hyperthermia at Duke was given using microwave applicators. Dielectrically loaded waveguides operating at 915 MHz (Clinitherm Mark IV, Texas) were used with either oil or water coupling. Some tumours were heated with spiral microwave applicators, operating at 433 MHz (Samulski et al. 1990). Surface cooling was used when the tumour did not involve skin, with bolus temperatures circulating at 40-41 "C. Thermometry was done using fiberoptic thermometers (Luxtron Corp., CA) with thermal mapping employed in tumours combined with fixed point skin surface thermometry (Engler et al. 1987). A mean of 17.5*4.8 temperature measurements were made in tumour and 5.9*2.9 points were measured in normal tissue. Patients returned home after completion of therapy. Patients returned for follow-up at 1, 3. 6, and 12 months after treatment and every 6 months thereafter. Tumours were categorized at each follow-up as complete response (no evidence of tumour) partial response ( > 50% decrease in tumour volume), no response (continued growth or < 50% decrease or no change), or recurrence (regrowth after a complete or partial response). Normal tissue damage was also evaluated at each follow-up. Histologic evaluation of normal soft tissues were performed on 32 dogs and bone changes were evaluated in 29 dogs at time of necropsy.
2 . 2 . Statistical methods The time from randomization until minimum tumour volume and the time from
312
S. McChesney Gillette et al.
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
randomization until tumour regrowth were analysed as censored survival data (Cox and Oakes 1984). Failure-time distributions are presented by plotting the Kaplan-Meier estimated survival distributions (Kaplan and Meier 1958, Cox and Oakes 1984). Both log rank and generalized Wilcoxon tests (Cox and Oakes 1984) were used to calculate the statistical significance of categorical variables, and both tests are reported. The difference between these two tests is that the log rank test weights each failure equally, while the latter test assigns more weight to earlier failures when there are more dogs at risk. Proportional hazards regression modelling (Cox and Oakes 1984) was used to assess the association of temperature data with time until minimum tumour volume and, separately, with time until tumour regrowth. Logistic regression (Cox 1970) was used to test for associations of temperature data with the risk of metastases. 2.3. Temperature distribution descriptors Several temperature distribution descriptors were used for this study. Descriptors were the maximum, median and the means of the T,s, T 5 0 ~for , each treatment. T,,s and T,s were calculated as previously described (Leopold et al. 1992). Treatment course evaluations also included the highest per treatment descriptor, the median of the per treatment descriptors and the mean of the two highest per treatment descriptors.
3. Results Sixty-four dogs were included in the study. Sixteen dogs were treated at NCSU/Duke and 48 were treated at CSU. Of these, 33 were randomized to receive XRT only and 3 1 to receive XRT and HT (Table I ) . Distribution by histologic type, stage, and volume is shown in Table 2. 3.1. Overall comparison of XRT and HT plus XRT An overall comparison of the two treatment arms revealed no difference in the frequency of complete responses (30.3% or 10/33 dogs for XRT alone and 41 ~ 9 % or 13/31 dogs for XRT plus HT). However, duration of local control was prolonged with XRT plus HT compared with XRT alone (Figure 1; Wilcoxon test p=0-046; log rank test p=O.O64). Median times to tumour regrowth were 530 days at CSU and 430 days at NCSU/Duke. The difference was not significant by log rank test p=0-0665 or Wilcoxon test p=0.750. There was a total of 14 late normal tissue complications; seven in each treatment arm. These consisted of soft tissue andlor bone necroses, with no obvious differences in the type or severity between the two arms. The difference in frequency of late radiation injuries was not statistically significant for the two treatment arms (Table 3). Histologically, changes were noted in the skin and oral mucous membranes of 32 dogs ( I 3 treated with HT plus XRT and 19 with XRT alone) that underwent full necropsies at the time of death. Types of lesions noted included epidermal and adnexal atrophy, dermal fibrosis and atrophy of oral accessory salivary glands. There was no difference in the frequency of these mild lesions between the two treatment arms (10/13,77% for HT plus XRT and 16/19, 84.2% for XRT alone). Soft tissue changes included mild muscle fibrosis, endoneurial and perineurial fibrosis and perivascular fibrosis and intimal proliferation within small arteries. The incidence of these lesions was similar for HT plus XRT (8/13, 62%) and XRT alone (14/19,74%) treatment arms as well. The most prominent feature of bone changes was the appearance of empty lacunae, indicating loss of parenchymal cells. However, the frequency of this event was not different for the two treatment arms (9/12, 75% for XRT plus HT and 10/15, 67% for XRT alone).
313
Heat plus irradiation of sarcornus Table 2. Tumour distribution by histologic type, stage, location, volume and grade. Radiation Radiation plus Tumour characteristic only hyperthermia Total Histologic type Hemangiopericytoma Fi brosarcoma Other
18 10 5
13 14 4
31 24 9
2 13 12 6
4 12 10 5
6 25 22 11
20 7 6 0
16 13 0 2
36 20 6 2
6 15 11 1
8 18 5 0
14 33 16
19 8 3 3
13 9 2 7
32 17 5 10
8
15 29
Stage TI
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
7.2 T3 T4 Location Extremity Oral Trunk Head and neck
Volume (cm2) < 10 10-99 100-210 Missing French Grade' 1 2 3 Missing
1
NCI Grade'
7 18 5 3
1
2 3 Missing
11
10
5 7
10
'Coinde c/ a/. 1988. 'Costa ef a/. 1984.
0.0
I
200
4b0
I
600
1
800
I
1000
1
1200
1
1400
r
1600
Figure I . Kaplan-Meier curves for estimates of recurrence time for radiation only (- - -) and radiation plus hypenhermia (-). Median time to recurrence was significantly increased with hyperthermia.
S. McChesney Gillette et al.
314
Table 3. Comparison of late effect frequencies for XRT and HT plus XRT. Radiation dose (Gy) 47.5 50.0 52.5 55.0 45.0 42.5 40.0 Treatment 35.0 XRT~ HT PIUS XRT'..'
013 115
011 016
-
I I5
011 215
014
315
215
215
315 -
111 -
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
'Frequency of late effects was correlated with increasing radiation dose (p=0.005). 'Frequency of late effects was not correlated with increasing radiation dose ( p = 0 . 6 0 ) . 'For these radiation doses used in both treatment arms in which complications were observed, the tendency toward a higher frequency at lower radiation doses for HT plus XRT group was not significant ( p = 0 . 7 6 ) .
There were no significant differences in the frequencies of distant metastases between the treatment arms at 1 or 2 years post treatment. Frequency of metastases were 7 . 4 % for radiation at 1 year and 20%for radiation plus hyperthermia. At 2 years after treatment the frequencies were 1 1 -5% for radiation only and 25% for radiation plus heat. 3.2. Patient related prognostic factors Several of the patient related prognostic factors showed significant correlations with time to minimum volume, including histology, location, grade and tumour volume (Table 4). Tumours on the extremities took longer to reach minimum volume than those of the trunk (p=O.O43) or head and neck (p=O.o04). Data are not shown. 3.3. Treutment related prognostic factors There was no correlation between radiation dose received and time to minimum tumour volume or probability for regrowth at 12 months post treatment, in either treatment arm (Table 5). In contrast, there was a positive correlation between XRT dose received and probability for late normal tissue damage, but only for the animals treated with XRT alone (Table 3). Thermal data on 29 dogs were analysed to determine 15 different descriptors of the normal or temperature distributions achieved during heating (Table 6). Thermometry on one dog was lost and, thus this animal was excluded from the thermal analysis. Nineteen dogs received ultrasound HT at CSU and 10 dogs received microwave HT at NCSU/Duke. Mean heating durations per treatment were 32.8 (4.3) min at CSU and 30.8 ( I ~ 7 min ) at NCSU/Duke. Median normal tissue temperature was 37.5"C with a range of 35.01-41.72"C. Median T, and T, were 37.79 and 34.78 for normal tissues. At CSU the mean number of thermometry points per treatment was 4 - 6 ( I -4) in tumour. At Duke, thermal mapping in tumours resulted in a larger number of tumour points than at CSU 17.5 (8.4), but the number of normal tissue points was similar to CSU 5 . 9 (2.9). Median tumour T,,s and T,'s were 41.3 (0.6) and 39.3 (0-5)"Cat Duke and 41 a 0 (0.2) and 37.8 (0.2)"C at CSU, respectively. Although the descriptors were slightly higher for NCSU/Duke animals than CSU animals, the differences were not statistically significant ( p Z 0 *09). Thirteen of 15 tumour temperature distribution descriptors shown in Table 6 were highly correlated with time to minimum volume ( p s 0 . 0 1 ) . The descriptors median T,, and the average of the two highest T,'s were not correlated. Minimum T,, was correlated with time to minimum tumour volumep=0.05 (Table 6). Normal tissue temperature descriptors were not correlated with time to minimum volume (data not shown). In contrast to the correlations seen with time to minimum volume, there were no significant correlations between time to local regrowth and any of the tumour or normal tissue temperature descriptors (Table 6). Temperature distribution data were also analysed for their correlations with normal tissue complications, but no significant relationships were found (data not
Heat
plus irradiation of sarcomas
315
Table 4. Prognostic variables and their relation to time to minimum tumour volume. Variable
Median dayst
Log rank Test
Wilcoxon Test
224 (133, 379) 116 (53, 287)
0.419
0.110
177 (79, 287) 224 (133. 379)
0.330
0.160
283 (192, 385) 113 (79, 172) 296 (108, 5 5 8 )
0.004
0.016
120 (21, 287) 260 (172, 396) 115 (79, 221) 280 (115, 03)
0.260
0.230
291 (219, 391) 43 (24, 59) 115 (102, 210)
0.005
0.001
224 (115, 392) 266 (108, 385) 56 (21, 79)
0.009
0.018
210 (113, 396) 226 (108, 385) 99 (52, 201)
0. 100
0.230
197 (49, 385) 269 (192, 392) 99 (52. . . 201)
0.300
0.050
Institution
csu
NCSU/Duke Treatment
Radiation alone Radiation plus heat Tutnour type'
HPC Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
FSA
Other Stage
TI T, T3 TJ Location2
Extremity Trunk Head plus neck French grade' 1
2 3 NCI grade 1
2 3 Tumour volume (cm')
0-9 10-98 99 or more
I
t95X confidence intervals are given in parentheses. 'FSA differs significantly from the other two histologic types (adjusted log rank p
1992,
VOL.
8,
NO. 3,
309-320
Response of canine soft tissue sarcomas to radiation or radiation plus hyperthermia: a randomized phase I1 study
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
S. McCHESNEY GILLETTEt II, M. W. DEWHIRSTg, E. L. GILLETTEt, D. E. THRALLI, R. L. PAGEI, B. E. POWERS?, S. J. WITHROW$, G. ROSNERS, C. WONG* and D. A. SIM* tDepartment of Radiology and Radiation Biology, and $Department of Clinical Sciences, Colorado State University, Fort Collins, Colorado 80523, USA $Division of Radiation Oncology, Duke University Medical Center, Durham, North Carolina 27710, USA fCollege of Veterinary Medicine, North Carolina State University, Raleigh, North Carolina 27606, USA *Clinical Trials Resources Group, John P. Robarts Research Institute, PO Box 5015, London Ontario, N6A 5KB Canada (Received 25 October 1991)
Sixty-four dogs with spontaneous soft tissue sarcomas without evidence of metastases were stratified by tumour volume and randomized to receive graded doses of radiotherapy (XRT) alone or radiotherapy plus hyperthermia (HT). An improvement in duration of local control was achieved with the addition of hyperthermia as compared with XRT alone (Wilcoxon, p=0.040; log rank, p=0.064). Overall frequency of late complications was not different for the two treatment arms when comparing across equivalent XRT dose groups. Frequency of distant metastases after therapy completion was not significantly different for the two treatment arms at 1 year (7.4% for XRT versus 20% for HT plus XRT) or 2 years ( I 1 . 5 % for XRT versus 25% for HT plus XRT) post therapy. These results suggest that a therapeutic gain was achieved for this group of turnour-bearing animals. Uni- and multivariate analyses were performed to examine the potential for various factors to influence treatment outcome. Patient related variables included tumour stage, histologic subtype and grade and tumour site. Treatment related variables included total radiation dose and 15 descriptors of temperature distributions achieved during hyperthermia. When considering parient related factors, tumour histology, grade and location were important predictors of time to minimum volume, but only turnour location influenced time to tumour regrowth. When considering treatmen/ related factors, radiation dose was not significantly correlated with time to minimum volume or time to local regrowth, but it was correlated with probability for late normal tissue damage in the XRT alone group (p=O.OOS). For the hyperthermia treatments, I3 of 15 tumour temperature distribution descriptors were correlated with time to minimum volume, but none were correlated with time to local regrowth. These results suggest that caution should be used in interpreting the value of temperature distribution descriptors in predicting for long-term local control after hyperthermia and radiotherapy. based on analysis of short-term responses. Key words: hyperthermia, radiation, canine sarcomas
1. Introduction Two Phase I1 studies using radiotherapy plus hyperthermia (HT) as a combined modality with graded doses of radiotherapy (XRT) or XRT plus surgery of human patients with soft tissue sarcomas have been published recently (Egawa et al. 1989, Leopold et al. 1989 and 1992). These studies evaluated the initial response to treatment as measured by changes IlCorrespondence to: Dr Sharon M. Gillette, Comparative Oncology Unit, Colorado State University, Fort Collins, Colorado 80523, USA 0265-6736/92 $3.00 01992 Taylor & Francis Lrd
S. McChesney Gillette et al.
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
310
in tumour volume and histologic parameters. In one study, four of 10 superficial tumours had a complete response to combined treatment (Egawa ef al. 1989). In a separate study, 34 patients with stage IKIB-IIA soft tissue sarcomas were randomized preoperatively to receive one or two fractions of HT per week, in conjunction with conventionally fractionated radiotherapy, to a total dose of 50 Gy (Leopold ef al. 1992). The complete response rate was relatively low in this patient population at 6 weeks after therapy completion. However, the group receiving two fractions of heat per week had a significantly greater percentage of patients with complete necrosis in surgically excised specimens than those patients receiving one fraction of heat per week. Furthermore, several descriptorsof the temperature distribution were positively correlated with percent necrosis (Oleson et al. 1989, Leopold et al. 1992). While these results suggest that HT can confer a therapeutic advantage in the treatment of soft tissue sarcomas, there continues to be a need for studies which address more relevent clinical endpoints, such as duration of local control, late normal damage and frequency of distant metastasis. Use of spontaneous tumours in pet animals has proven to be a good model system for testing the efficacy of new treatment modalities. Several studies using HT alone or combined with XRT have been done using spontaneous tumours (Marmor et al. 1978, Gillette et al. 1987, Dewhirst et al. 1983 and 1985, Brewer and Turrel 1982, Richardson et al. 1984). Three studies specifically evaluated canine soft tissue sarcomas (Dewhirst et al. 1983, Brewer and Turrel 1982, Richardson et al. 1984). All of the studies suggested an improvement in local control rates over that achievable with radiation alone, but small numbers of animals were accrued and thermometry was limited. In the present study, long-term follow-up was done on 64 spontaneous canine soft tissue sarcomas randomized to receive XRT alone or combined with HT with multipoint thermometry. This was a cooperative study between the Comparative Oncology Unit at Colorado State University (CSU) and the North Carolina Animal Cancer Treatment Program at North Carolina State University (NCSU). Hyperthermia treatments for NCSU dogs were done at Duke University Medical Center (DUMC). The emphasis of this analysis was to establish whether HT conferred any benefit over XRT alone, using short-term tumour regression, long-term local control, metastases and late normal tissue damage as endpoints. These endpoints were also evaluated for their correlation with patient and treatment related prognostic factors. 2. Materials and methods 2.1 . Experimental design Sixty-six dogs with soft tissue sarcomas were admitted to CSU or NCSU. All dogs were screened for metastatic disease by physical exam and chest radiographs. Results of a history, physical exam, CBC and blood chemistry panel were used to determine whether each animal was healthy enough to withstand treatment, which requires multiple anaesthesia procedures. Histologic type was confirmed by biopsy on all tumours before entry into the study. Two grading schemes were used to classify the tumours (Costa et al. 1984, Coindre et al. 1988). Criteria for exclusion included tumour volume >250 cm’, tumour depth > 4 0 cm, tumour margins larger than available hyperthermia applicator size, prior radiotherapy, chemotherapy or hyperthermia to the study lesion, and evidence of distance metastases. Tumours were also staged according to World Health Organization criteria (T, < 2 cm diameters, T, 2-5 cm diameter, T,>5 cm diameter or invasion of subcutis, T, tumour invading other structures) (WHO 1979). Two dogs were excluded from the final analysis. One patient was omitted because of thermal toxicity that occurred during the first hyperthermia treatment which necessitated amputation. A second patient was omitted because it developed metastases before the end of treatment. A third tumour was
-
Heat plus irradiation of sarcomas
311
Table 1. Dogs randomized into the study. Radiation Dose Radiation plus heat GY) only 5 6 5
47.5 50.0
3 7 0 7 4 5
52.5 55.0
5
2
0 0
33
31
35.0 40.0 42.5 45.0
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
Total
5 5 5
not considered because it was the second of two tumours occurring on the same animal. Dogs were stratified by tumour volume and randomized to receive total radiation doses between 35 and 55 Gy with or without hyperthermia (Table 1). Radiation was delivered in 10 fractions on a Monday, Wednesday and Friday schedule lasting 22 days. At CSU dogs were irradiated using a clinical linear accelerator which produced 6 MV X-rays. At NCSU a 6oCoteletherapy unit was used for irradiation. All irradiation and hyperthermia treatments were done under general anaesthesia using thiopental for induction and isoflurane for maintenance. Treatment planning using an Artronix treatment planning computer was done at both institutions. The prescribed tumour dose was defined as the minimum tumour dose. Maximum tumour dose never exceeded minimum dose by more than 15% . A 2-cm margin of normal tissue was included in the treatment volume. Hyperthermia was given approximately 3 h following irradiation. The target temperature was a minimum tumour temperature of 42°C without exceeding 40°C in normal tissues. At CSU hyperthermia was delivered with a 2 a3 MHz ultrasound transducer. Thermometry was done using a total of 10 thermistors in tumour and normal tissues (Gillette et al. 1987). Duration of HT was 30 min from the time that any intratumoral temperature reached 42°C. HT treatments were given twice weekly on a Monday-Friday schedule. At CSU the mean number of thermocouples was 4 . 6 & 1 - 4 in normal tissues and 5 - 4 + 1.4 in tumour. Hyperthermia at Duke was given using microwave applicators. Dielectrically loaded waveguides operating at 915 MHz (Clinitherm Mark IV, Texas) were used with either oil or water coupling. Some tumours were heated with spiral microwave applicators, operating at 433 MHz (Samulski et al. 1990). Surface cooling was used when the tumour did not involve skin, with bolus temperatures circulating at 40-41 "C. Thermometry was done using fiberoptic thermometers (Luxtron Corp., CA) with thermal mapping employed in tumours combined with fixed point skin surface thermometry (Engler et al. 1987). A mean of 17.5*4.8 temperature measurements were made in tumour and 5.9*2.9 points were measured in normal tissue. Patients returned home after completion of therapy. Patients returned for follow-up at 1, 3. 6, and 12 months after treatment and every 6 months thereafter. Tumours were categorized at each follow-up as complete response (no evidence of tumour) partial response ( > 50% decrease in tumour volume), no response (continued growth or < 50% decrease or no change), or recurrence (regrowth after a complete or partial response). Normal tissue damage was also evaluated at each follow-up. Histologic evaluation of normal soft tissues were performed on 32 dogs and bone changes were evaluated in 29 dogs at time of necropsy.
2 . 2 . Statistical methods The time from randomization until minimum tumour volume and the time from
312
S. McChesney Gillette et al.
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
randomization until tumour regrowth were analysed as censored survival data (Cox and Oakes 1984). Failure-time distributions are presented by plotting the Kaplan-Meier estimated survival distributions (Kaplan and Meier 1958, Cox and Oakes 1984). Both log rank and generalized Wilcoxon tests (Cox and Oakes 1984) were used to calculate the statistical significance of categorical variables, and both tests are reported. The difference between these two tests is that the log rank test weights each failure equally, while the latter test assigns more weight to earlier failures when there are more dogs at risk. Proportional hazards regression modelling (Cox and Oakes 1984) was used to assess the association of temperature data with time until minimum tumour volume and, separately, with time until tumour regrowth. Logistic regression (Cox 1970) was used to test for associations of temperature data with the risk of metastases. 2.3. Temperature distribution descriptors Several temperature distribution descriptors were used for this study. Descriptors were the maximum, median and the means of the T,s, T 5 0 ~for , each treatment. T,,s and T,s were calculated as previously described (Leopold et al. 1992). Treatment course evaluations also included the highest per treatment descriptor, the median of the per treatment descriptors and the mean of the two highest per treatment descriptors.
3. Results Sixty-four dogs were included in the study. Sixteen dogs were treated at NCSU/Duke and 48 were treated at CSU. Of these, 33 were randomized to receive XRT only and 3 1 to receive XRT and HT (Table I ) . Distribution by histologic type, stage, and volume is shown in Table 2. 3.1. Overall comparison of XRT and HT plus XRT An overall comparison of the two treatment arms revealed no difference in the frequency of complete responses (30.3% or 10/33 dogs for XRT alone and 41 ~ 9 % or 13/31 dogs for XRT plus HT). However, duration of local control was prolonged with XRT plus HT compared with XRT alone (Figure 1; Wilcoxon test p=0-046; log rank test p=O.O64). Median times to tumour regrowth were 530 days at CSU and 430 days at NCSU/Duke. The difference was not significant by log rank test p=0-0665 or Wilcoxon test p=0.750. There was a total of 14 late normal tissue complications; seven in each treatment arm. These consisted of soft tissue andlor bone necroses, with no obvious differences in the type or severity between the two arms. The difference in frequency of late radiation injuries was not statistically significant for the two treatment arms (Table 3). Histologically, changes were noted in the skin and oral mucous membranes of 32 dogs ( I 3 treated with HT plus XRT and 19 with XRT alone) that underwent full necropsies at the time of death. Types of lesions noted included epidermal and adnexal atrophy, dermal fibrosis and atrophy of oral accessory salivary glands. There was no difference in the frequency of these mild lesions between the two treatment arms (10/13,77% for HT plus XRT and 16/19, 84.2% for XRT alone). Soft tissue changes included mild muscle fibrosis, endoneurial and perineurial fibrosis and perivascular fibrosis and intimal proliferation within small arteries. The incidence of these lesions was similar for HT plus XRT (8/13, 62%) and XRT alone (14/19,74%) treatment arms as well. The most prominent feature of bone changes was the appearance of empty lacunae, indicating loss of parenchymal cells. However, the frequency of this event was not different for the two treatment arms (9/12, 75% for XRT plus HT and 10/15, 67% for XRT alone).
313
Heat plus irradiation of sarcornus Table 2. Tumour distribution by histologic type, stage, location, volume and grade. Radiation Radiation plus Tumour characteristic only hyperthermia Total Histologic type Hemangiopericytoma Fi brosarcoma Other
18 10 5
13 14 4
31 24 9
2 13 12 6
4 12 10 5
6 25 22 11
20 7 6 0
16 13 0 2
36 20 6 2
6 15 11 1
8 18 5 0
14 33 16
19 8 3 3
13 9 2 7
32 17 5 10
8
15 29
Stage TI
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
7.2 T3 T4 Location Extremity Oral Trunk Head and neck
Volume (cm2) < 10 10-99 100-210 Missing French Grade' 1 2 3 Missing
1
NCI Grade'
7 18 5 3
1
2 3 Missing
11
10
5 7
10
'Coinde c/ a/. 1988. 'Costa ef a/. 1984.
0.0
I
200
4b0
I
600
1
800
I
1000
1
1200
1
1400
r
1600
Figure I . Kaplan-Meier curves for estimates of recurrence time for radiation only (- - -) and radiation plus hypenhermia (-). Median time to recurrence was significantly increased with hyperthermia.
S. McChesney Gillette et al.
314
Table 3. Comparison of late effect frequencies for XRT and HT plus XRT. Radiation dose (Gy) 47.5 50.0 52.5 55.0 45.0 42.5 40.0 Treatment 35.0 XRT~ HT PIUS XRT'..'
013 115
011 016
-
I I5
011 215
014
315
215
215
315 -
111 -
Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
'Frequency of late effects was correlated with increasing radiation dose (p=0.005). 'Frequency of late effects was not correlated with increasing radiation dose ( p = 0 . 6 0 ) . 'For these radiation doses used in both treatment arms in which complications were observed, the tendency toward a higher frequency at lower radiation doses for HT plus XRT group was not significant ( p = 0 . 7 6 ) .
There were no significant differences in the frequencies of distant metastases between the treatment arms at 1 or 2 years post treatment. Frequency of metastases were 7 . 4 % for radiation at 1 year and 20%for radiation plus hyperthermia. At 2 years after treatment the frequencies were 1 1 -5% for radiation only and 25% for radiation plus heat. 3.2. Patient related prognostic factors Several of the patient related prognostic factors showed significant correlations with time to minimum volume, including histology, location, grade and tumour volume (Table 4). Tumours on the extremities took longer to reach minimum volume than those of the trunk (p=O.O43) or head and neck (p=O.o04). Data are not shown. 3.3. Treutment related prognostic factors There was no correlation between radiation dose received and time to minimum tumour volume or probability for regrowth at 12 months post treatment, in either treatment arm (Table 5). In contrast, there was a positive correlation between XRT dose received and probability for late normal tissue damage, but only for the animals treated with XRT alone (Table 3). Thermal data on 29 dogs were analysed to determine 15 different descriptors of the normal or temperature distributions achieved during heating (Table 6). Thermometry on one dog was lost and, thus this animal was excluded from the thermal analysis. Nineteen dogs received ultrasound HT at CSU and 10 dogs received microwave HT at NCSU/Duke. Mean heating durations per treatment were 32.8 (4.3) min at CSU and 30.8 ( I ~ 7 min ) at NCSU/Duke. Median normal tissue temperature was 37.5"C with a range of 35.01-41.72"C. Median T, and T, were 37.79 and 34.78 for normal tissues. At CSU the mean number of thermometry points per treatment was 4 - 6 ( I -4) in tumour. At Duke, thermal mapping in tumours resulted in a larger number of tumour points than at CSU 17.5 (8.4), but the number of normal tissue points was similar to CSU 5 . 9 (2.9). Median tumour T,,s and T,'s were 41.3 (0.6) and 39.3 (0-5)"Cat Duke and 41 a 0 (0.2) and 37.8 (0.2)"C at CSU, respectively. Although the descriptors were slightly higher for NCSU/Duke animals than CSU animals, the differences were not statistically significant ( p Z 0 *09). Thirteen of 15 tumour temperature distribution descriptors shown in Table 6 were highly correlated with time to minimum volume ( p s 0 . 0 1 ) . The descriptors median T,, and the average of the two highest T,'s were not correlated. Minimum T,, was correlated with time to minimum tumour volumep=0.05 (Table 6). Normal tissue temperature descriptors were not correlated with time to minimum volume (data not shown). In contrast to the correlations seen with time to minimum volume, there were no significant correlations between time to local regrowth and any of the tumour or normal tissue temperature descriptors (Table 6). Temperature distribution data were also analysed for their correlations with normal tissue complications, but no significant relationships were found (data not
Heat
plus irradiation of sarcomas
315
Table 4. Prognostic variables and their relation to time to minimum tumour volume. Variable
Median dayst
Log rank Test
Wilcoxon Test
224 (133, 379) 116 (53, 287)
0.419
0.110
177 (79, 287) 224 (133. 379)
0.330
0.160
283 (192, 385) 113 (79, 172) 296 (108, 5 5 8 )
0.004
0.016
120 (21, 287) 260 (172, 396) 115 (79, 221) 280 (115, 03)
0.260
0.230
291 (219, 391) 43 (24, 59) 115 (102, 210)
0.005
0.001
224 (115, 392) 266 (108, 385) 56 (21, 79)
0.009
0.018
210 (113, 396) 226 (108, 385) 99 (52, 201)
0. 100
0.230
197 (49, 385) 269 (192, 392) 99 (52. . . 201)
0.300
0.050
Institution
csu
NCSU/Duke Treatment
Radiation alone Radiation plus heat Tutnour type'
HPC Int J Hyperthermia Downloaded from informahealthcare.com by McMaster University on 02/14/15 For personal use only.
FSA
Other Stage
TI T, T3 TJ Location2
Extremity Trunk Head plus neck French grade' 1
2 3 NCI grade 1
2 3 Tumour volume (cm')
0-9 10-98 99 or more
I
t95X confidence intervals are given in parentheses. 'FSA differs significantly from the other two histologic types (adjusted log rank p
