- Email: [email protected]
Impact of biological clearance on tumor radioresponsiveness
Int. J. Radmtion
Oncology
Biol.
Phys.. Vol. 34. No. 2, pp. 389-39.X 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved 0360.3016/96 $1.5.00 + .OO
ELSEVIER
0360-3016(95)00222-7
l
Biology Original Contribution IMPACT
OF BIOLOGICAL
KIYOSHI
CLEARANCE
ON TUMOR
RADIORESPONSIVENESS
OHARA, M.D.,*,+ YOSHINORI HAYAKAWA, PH.D.,* HIROSHI HIDEO TATSUZAKI, M.D.*,+ AND YUJI ITAI, M.D.**+
FUJI, M.D.,*
*Department of Radiology, University Hospital, ‘Institute of Clinical Medicine and *Institute of Basic Medical Sciences, University of Tsukuba, Tsukuba City, Japan Purpose: To determine the capacity of biological clearance in tumor regression following radiotherapy by using metastatic brain tumors as a clinical model in which mechanical clearance is negligible. Methods and Materials: Thirty-eight tumors (19 nonsmall cell lung cancer, 11 small cell lung cancer, and 8 nonlung cancer) in 23 patients were followed with computed tomography (CT) scans over 3 months or more following initiation of radiotherapy, with doses ranging between 34 and 66 Gy. The tumor regression rate (RR; mm3/day), which represented the capacity of biological clearance, was calculated for each CT observation period. The complete response (CR) rate was calculated. The relationship between RR and tumor diameter was determined with regression analysis in conjunction with the pattern of contrast enbancement and the type of primary disease. The change of the RR also was examined. Results: The CR rate was 60.5% for the total group; it was lower for ring-enhanced tumors (41.7%) than wy enhanced tumors (69.2%), which included mostly small cell lung cancer metastases. The RR correlated significantly with the tumor diameter (D), with a regression curve of exponential function (RR = 0.035 *D’.§). Tbe RR varied widely and was rather large until 40 days following initiation of radiotherapy, especially for the subgroups of diffusely enhanced tumors and the small cell lung cancer tumors, and became rather constant thereafter. Conclusion: A tumor diameter exponent in the regression curve of smaller than 3.0 indicates that the larger the tumor volume is, the smaller the capacity of biological clearance. The capacity of biological clearance also is dependent on vascularity and cellularity of the tumor components expressed by the pattern of contrast enhancement. Tumor regression, Radioresponsiveness,
Biological clearance, Metastatic brain tumors.
INTRODUCTION
a tumor mass following treatment does not necessarily indicate the presence of viable tumor cells. Instead, the rate of tumor regression following sterilization of the entire clonogenic tumor cell population is determined primarily by the capacity of clearance of these oncologically inactivated components. This involves two main mechanisms; that is, mechanical ckarunce, such as exfoliation and shedding, and biological clearance, such as phagocytic scavenging and liquid permeation. Both
The radioresponsiveness of a tumor has been demonstrated to be of clinical use as a convenient and potent predictor of local control in radiotherapy, especially for head and neck tumors (1,4, 15) and uterine cervical cancer (11, 12, 16). The complete disappearance of tumor following radiotherapy is highly suggestive of a successful outcome, although it is controversial whether rapid initial regression is a prerequisite for this (2, 18). However, complete response is not a perfect predictor because of its rather high false positivity (20); the presence of persistent tumor, immediately following completion of radiotherapy, does not necessarily predict recurrence. Patients with a false positive persistent tumor may be needlessly managed with needless supplementary treatments that are often disabling. In other words, the presence of
mechanisms are involved in tumors on an outer tissue surface, such as head and neck tumors and uterine cervical cancer. However,
only biological
clearance can serve in
tumors completely surrounded by normal tissues that block mechanical clearance. An example of this is a metastatic brain tumor, which has the advantage of being ame-
nable to the assessment of size and texture via computed tomography
Reprints requests to: Kiyoshi Ohara, M.D., Institute of Clinical Medicine, University of Tsukuba, Tsukuba City 305, Japan.
(CT). The purpose of this study is to deter-
Accepted for publication 12 May 1995. 389
390
I. J. Radiation
Oncology
0 Biology
0 Physics
mine the impact of biological clearance on tumor regression using metastatic brain tumors as a clinical model.
METHODS
AND MATERIALS
Sequential CT scans of 23 patients with one or more metastatic brain tumor(s) treated by radiotherapy were evaluated. These CT scansmet the following criteria: (a) a follow-up period of 3 months or more, and a scanning frequency of twice or more following the initiation of radiotherapy; (b) use of contrast enhancement; and (c) tumors treated with a total prescribed dose of at least 30 Gy. The CT scans were obtained with several different types of CT scannersduring a period between 1983 and 1993, but all were obtained with 1 cm slice thickness. The CT follow-up periods ranged from 94 days to 7 years. Eight patients were women and 15 were men, with agesranging between 34 and 88 years. Five patients had a single metastasisand 18 patients had multiple metastases, among which the one or two largest measurabletumors were chosen for the study. Thus, 38 tumors were included: 19 tumors in 12 nonsmall cell lung cancer (NSCLC) patients, 11 tumors in 6 small cell lung cancer (SCLC) patients, 6 tumors in 4 breast cancer patients, and 2 tumors in 1 colon cancer patient. The tumors were treated either with “Co gamma-rays or with 6 MV x-rays, with whole brain irradiation, focal brain irradiation, or both, basically with a weekly dose of 9.0-10.0 Gy/5 fractions. The total prescribed doses ranged between 34.3 Gy/20 fractions and 65.8 Gyl38 fractions: doseslessthan 40.0 Gy for 6 tumors, between 40.0 and 50.0 Gy for 10 tumors, and greater than 50.0 Gy for 22 tumors. In most cases,corticosteroids and/or osmotic diuretics were administered prior to and during the course of radiotherapy. Ommaya’s reservoir was surgically placed for five tumors, three prior to and two during the course of radiotherapy, to reduce tumor masseffect by draining out part of the liquid component of the tumor. Most patients underwent chemotherapy, with a variety of cytotoxic agentsand/or hormone therapy, concurrently or following brain irradiation, for the managementof extracranial disease(s). Tumors were qualitatively classified into two subgroups, basedon the pattern of contrast enhancement: (a) ring enhanced,with central low density; and (b) diffusely enhanced, containing small amounts of low density. The presence of peritumoral edema also was assessed.The volume of each tumor was estimated as a sphere with diameter calculated as the average of the maximum and the minimum diameter on a tumor-center cross-section using two dimensional planimetry; most of the tumors were actually spherical. The complete response(CR) rate was calculated using the Kaplan-Meier method. The tumor regression rate (RR), a daily tumor volume decrement (r&/day), was calculated for each observation interval, and group data was computed based on initial tumor diameter, pattern of contrast enhancement, and the
Volume
34,
Number 2, 1996
type of primary disease (NSCLC, SCLC, and nonlung cancer (non-LC)). The change of RR in conjunction with the time elapsed following initiation of radiotherapy also was examined.
RESULTS Twelve tumors (31.6%) were ring-enhanced and 26 tumors (68.4%) were diffusely enhanced. The pattern of contrast enhancement did not change during the observation period, but calcifications appeared in three tumors. The percentageof ring-enhanced tumors in each subgroup (NSCLC, SCLC, and non-LC) was 36.8% (7 out of 19), 18.2% (2 out of 1l), and 37.5% (3 out of S), respectively. Peritumoral edema was observed in 28 tumors (73.7%) prior to radiotherapy. The tumor volume prior to radiotherapy ranged between 113 mm3 and 83,325 rnrr?, and that of five reservoir-placed tumors ranged between 17,149 mm3 and 83,325 &. The mean volume of the ring-enhanced tumors was significantly larger than that of the diffusely enhanced (16,559.2 mn? vs. 3,623.6 mms, p = 0.0017). All tumors decreasedfollowing radiotherapy with five tumors showing a transient increase (Fig. 1). Twentythree tumors (60.5%), including three calcified tumors, attained CR without any peritumoral edema between 28 days and 27.5 months following initiation of radiotherapy: 5 tumors (41.7%) were ring-enhanced and 18 (69.2%) were diffusely enhanced. The cumulative CR rates at 60, 90, and 120 days following initiation of radiotherapy were 18.7%, 30.0%, and 49.3%, respectively. Those for the diffusely enhanced were 24.0%, 37.3%, and 76.1%, respectively. Fifteen tumors failed to attain CR. Nine of these, however, converted to low density, non-enhancing lesions without peritumoral edema, while the remaining six continued to enhance with contrast. Five tumors recurred by the period between 4 and 10 months following
-30
0
30
ho
w
120 IS0
I x0
Days after RT initiation
Fig. 1. Change of the tumor volume following radiotherapy. The open circles indicate the ring-enhanced tumor and the closed circles the diffusely enhanced tumor. The arrow indicates the placement of Ommaya’s reservoir.
Biological
3.5
. I. *. Y = 2.49X I.-t6
- -,
*.
tumor
clearance
0 K. OHARA
391
et al.
‘.
3
00
.6 .7
.x .9
I
1.1 1.2 I.3 I.4 I.5 1.6 1.7
0
log(x) of tumordiameter(mm)
-
20
.
30
8
60
.
80
Days after
.
’
.
’
4
100 120 l-10 160 180 KT initiation
Fig. 2. Relationshipbetween the tumor regressionrate (RR: mm?/day)and the tumor diameterof entire data. The opencirclesindicatethe ring-enhancedtumor andthe closedcirclesthe diffusely enhancedtumor. The line indicates the regression curve with all tumors.
Fig. 3. Relationshipbetweenthe ratio of tumor regressionrate (RR (observed/expected))andthe time elapsedfollowing initiation of radiotherapy (entire data). The open circles indicate the ring-enhancedtumor and the closedcircles the diffusely enhancedtumor.
treatment; four showed contrast enhancement,
RR (expected) calculated with Eq. 1 and the time elapsed following initiation of radiotherapy was analyzed. The midpoint of each observation period was used as the elapsed time. Those RR data of zero or negative value were included for this analysis. The RR (O/E) varied widely and was rather large until approximately 40 days following initiation of radiotherapy, especially for the diffusely enhanced, and then became nearly constant, showing a biphasic distribution (Fig. 3). There was no significant correlation between them. Therefore, we divided the data into two subgroups based on elapsed time (140 days and >40 days), and again studied the relation between the RR and the tumor diameter. There was a significant correlation between the RR and the tumor diameter (Table 1); however, no significant correlation was observed between the RR (O/E) and the elapsed time.
and only
one had previously attained CR. The following data were excluded from calculation of RR: (a) zero or negative RR values; (b) interval between consecutive scans longer than three months; (c) residual calcification only; and (d) affected by drainage. The RR significantly correlated with tumor diameter (mm) when plotted on a dual logarithmic scale (Fig. 2). With regression analysis the regression curve of the RR was derived as an exponential function of the tumor diameter: A = 0.0347*D2.49,
(Eq. 1)
where A is the RR and D is tumor diameter. An exponential function is normally used in allometry: measuring the growth of a part in relation to a whole living organism. The RR for each subgroup significantly correlated likewise, with the constant and the exponent of the equation shown in Table 1. To examine the chronologic change of the RR following initiation of radiotherapy, a relationship between the RR (O/E); that is, the ratio of the RR (observed) to the
DISCUSSION Radioresponsiveness is an expression not only of radiosensitivity, or a decrease of the tumor cell population, but of a total balance of the decrease and increase in all tumor
Table 1. Constantand exponentof regressioncurve in relationshipbetweentumor regressionrate and tumor diameter Subgroup
Data no.
Constant(c)
Exponent (E)
R
p-Value
Entire Ring-enhanced Diffusely-enhanced NSCLC SCLC Non-LC 5 40 days > 40 days
66 24 42 36 18 12 35 31
0.0347 0.0123 0.0316 0.0407 0.0347 0.0199 0.0276 0.0598
2.49 2.77 2.58 2.41 2.55 2.61 2.63 2.21
0.88 0.84 0.92 0.87 0.85 0.85 0.89 0.87
0.0001 0.0001 0.0001 0.0001 0.0001 0.0004 0.0001 0.0001
The regressioncurve: A = C*D’, whereA is the tumor regressionrate (mrn’/day) in an observationinterval and D is the initial tumor diameter(mm’) in this interval; R: a correlationcoefficient. NSCLC: nonsmallcell lung cancer,SCLC: smallcell lung cancer, non-LC: nonlungcancer, 5 40 days: observationpoint at < 40 days, and > 40 days: at > 40 days.
392
I. J. Radiation
Oncology
0 Biology
l Physics
components. These include (a) cellular components: tumor cells that are proliferative, quiescent, or sterilized, stromal cells, infiltrating cells such as inflammatory cells and macrophages; and (b) organic stromal components: tumor-produced stromal tissues, radiation-induced fibrous tissues, and granulation tissues; and (c) inorganic components: tumor products such as keratin and mucin, necrotic tissues coagulated or liquified, inflammatory and edematous exudates, and so forth. The distribution of these components in tumors will vary with many factors including tumor size, histologic type, and tumor environment (especially vasculature of the tumor and the surrounding normal tissues). In exophytic tumors facing outward, some of these components can be removed by both mechanical and biological clearance; therefore, their individual capacities can not be measured separately. However, in tumors isolated from the outside, all tumor components are removed by biological clearance alone. Accordingly, its capacity can be measured as the RR. If the capacity of biological clearance is proportionate to tumor volume (D’), and tumors are qualitatively identical, the slope of tumor volume regression curves should be identical to each other independent of the tumor volume prior to radiotherapy (5). The slope of the curves obtained, however, was steeper for small tumors than for large tumors. This was shown more clearly by the fact that the RR significantly correlated with the tumor diameter with an exponent of 2.49 of the regression curve. Although the value of the exponent and the constant may be influenced by one another, the exponents of all subgroups were less than 3.0. This indicates that the smaller the tumor volume is, the larger the RR. Moreover, it was as small as 2.21 for the subgroup of 240 days, suggesting that the RR is proportionate to the tumor surface area (0’). This would suggest that biological clearance arises mainly at the peripheral zone of the tumor; that is, the tumor-normal tissue interface. The RR (O/E) values varied widely and were rather large in the initial phase of 40 days following radiotherapy, particularly for the diffusely enhanced tumors. Actually, the diffusely enhanced tumors attained CR more rapidly and to a higher rate than the ring-enhanced tumors. The higher initial RR (O/E) value of the diffusely enhanced tumors is due partly to the fact that this group included most of the data from the subgroup of SCLC, a tumor known to be more radiosensitive than the others. This higher initial rate has been observed also by Kataoka et al., although the type of primary disease was not described (13). The RR (O/E), however, did not correlate with the elapsed time following initiation of radiotherapy. This lack of correlation may be due to the short and irregular observation period of up to 4 months and the small number of patients. However, the fact that a tumor mass of low density without contrast enhancement persisted for 8 months could support the biphasic response in tumor regression. Bataini et al. have reported the biphasic distribution in
Volume
34, Number
2, 1996
tumor regression (3). The change of the CR rate of metastatic cervical lymph nodes from primary disease of the oropharynx and pharyngolarynx showed an initial steep linear phase followed by a period of constant decrease in the slope. The CR rate was significantly lower for lymph nodes than for the primary disease at the end of radiotherapy, but the percentage of cases for each group with residual disease were almost equal 2 months later. This biphasic response was also observed for some salivary gland tumors, which were treated with neutrons (5). The pattern of tumor regression of prostate cancer reported by Stmad et al. strongly suggests a biphasic response as well (19). While these tumors were all isolated from outer surfaces, no qualitative assessment was made. One hypothesis for the biphasic response, proposed by Bataini et al. suggests that radiation-induced impairment of blood flow reduces the clearance rate (3). This hypothesis is supported by the fact that tumor regression is prolonged by radiation-induced stromal cell damage that was shown experimentally with a model of the severe combined immunodeficient mouse (6). Bessel1 et al. have explained the cause for the biphasic response as follows: The regression depends on the ratio of proliferating cells to stroma in which the initial regression represents the rapid clearance of the dead cellular component, followed by the slow clearance of the stromal component (5). In metastatic brain tumors, the area of central low density is considered to be composed of biologically inactivated components, mostly of liquified necrosis and possibly exudates, because they were readily drained through a reservoir in large tumors. In this context, the ring-enhanced tumors contain rather small amounts of vital components of tumor cells and tumor stromal tissues, which exist only at the peripheral rim of the tumor mass. Conversely, the diffusely enhanced tumors consist largely of vital components distributed throughout the tumor mass. These enhancement patterns suggest alternate explanations for the biphasic response. The first is that the tumors in the initial phase contain a proportionately larger amount of vital cellular components, especially in the diffusely enhanced tumors, which decrease volume by cytolysis and demonstrate cell volume reduction with interphase cell death or apoptosis (8,17) induced by irradiation. The second is that the tumor cells that are well nourished by the functional stroma should be more radiosensitive. The third is that biologic clearance can arise, even deep within a diffusely enhanced tumor, through the functional tumor stroma. The capacity of biological clearance can differ with the type and conditions of normal tissues in which the tumor has arisen. In experimental studies comparing the rate of dead tumor cell removal from brain, muscle, and SC tissues of rats, it is most rapid in muscle and slowest in brain (14). This difference may be due to qualitative and quantitative differences in these tissues including the stroma and scavenging cells, which are microglia and astrocyte in the brain (10). Moreover, the rate of biological clearance in these
Biological
tumor
clearance
normal tissues can be affected by irradiation, probably in a dose-dependent fashion. Cairncross et al. have reported histopathologic findings of me&static brain tumors that persisted for months following radiotherapy (7). The tumor mass lesions showed discrete, well-circumscribed regions of necrosis surrounded by gliosis without tumor cells. Our nine tumors, which persisted as low-density masses without contrast enhancement or petitumoral edema, may have been this type of sterilized tumor. Magnetic resonance imaging (MRI), which provides more information about tumor quality, has demonstrated large endophytic uterine cervical tumors that persisted at 6 months following radiotherapy and resolved at 9 months (9). These tumors with slow regression showed declining signal intensity of the solid component, which probably represents inor-
0 K. OHARA
et al.
393
ganic elements such as coagulated necrosis and fibrous tissues. The solid component may be more slowly removed than the liquid component, because permeation clearance can also serve for the latter. Thus, the tumor quality following radiotherapy is important to predict local tumor control. In conclusion, evaluation of the RR of metastatic brain tumors following radiotherapy reveals clearance as a function of an intermediate range between the tumor volume (D’) and the tumor surface area (D’). Therefore, the larger the tumor volume, the smaller the capacity of biological clearance. In addition, this is larger in tumors with less necrosis than in tumors with a large amount of necrosis, particularly in the early phase following initiation of radiotherapy.
REFERENCES 1. Barkley, H. T., Jr.; Fletcher, G. H. The significance of residual disease after external irradiation of squamous-cell carcinoma of the oropharynx. Radiology 124:493-495; 1977. 2. Bartelink, H. Prognostic value of the regression rate of neck node metastases during radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 9:993-996; 1983. 3. Bataini, J. P.; Bemier, J.; Jaulerry, C.; Brunin, F.; Pontvert, D.; Lave, C. Impact of neck node radioresponsiveness on the regional control probability in patients with oropharynx and pharyngolarynx cancers managed by definitive radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 13:817-824; 1987. 4. Bataini, J. P.; Jaulerry, C.; Brunin, F.; Ponvert, D.; Ghossein, N. A. Significance and therapeutic implications of tumor regression following radiotherapy in patients treated for squamous cell carcinoma of the oropharynx and pharyngolarynx. Head Neck 12:41-49; 1990. 5. Bessel, E. M.; Catterall, M. The regression of tumors of the head and neck treated with neutrons. Int. J. Radiat. Oncol. Biol. Phys. 9:799-807; 1983. 6. Budach, W.; Taghian, A.; Freeman, J.; Gioioso, D.; Suit, H. D. Impact of stromal sensitivity on radiation response of tumors. J. Natl. Cancer Inst. 85:988-993; 1993. 7. Caimcross, J. G.; Chemik, N. L.; Kim, J.; Posner J. B. Sterilization of cerebral metastases by radiation therapy. Neurology 29:1195-1202; 1979. 8. Falkvoll, K. H. The relationship between changes in tumor volume, tumor cell density and parenchymal cord radius in a human melanoma xenograft exposed to single dose irradiation. Acta Oncol. 29:935-939; 1990. 9. Flueckiger, F.; Ebner, F.; Poschauko, H.; Tamussino, K.; Einspieler, R.; Ranner, G. Cervical cancers: Serial MR imaging before and after primary radiation therapy: A 2-year follow-up study. Radiology 184:89-93; 1992. 10. Gueneau, G.; Drouet, J.; Privat, A; Court, L. Differential radiosensitivity of neurons and neuroglia of the hippocam-
11.
12.
13.
14.
15. 16. 17. 18.
19.
20.
pus in the adult rabbit. Acta Neuropathol. 48:199-209; 1979. Hardt, N.; van Nagel, J. R.; Hanson, M.; Donaldson, E.; Yoneda, J.; Maruyama, Y. Radiation-induced tumor regression as a prognostic factor in patients with invasive cervical cancer. Cancer 49:35-39; 1982. Jacobs, A. J.; Faris, C.; Perez, C. A.; Kao, M.; Galakatos, A.; Camel, H. A. Short-term persistence of carcinoma of the uterine cervix after radiation: An indicator of long-term prognosis. Cancer 57:944-950; 1986. Kataoka, M.; Inatsuki, S.; Kawamura, M.; Itoh, H.; Hamamoto, K. Treatment response on CT scan in patients with brain metastases treated with irradiation. Radiat. Med. 9:41-46; 1991. Kumar, R. V.; Hoshino, T.; Wheeler, K. T.; Barker, M.; Wilson, C. B. Comparative rates of dead tumor cell removal from brain, muscle, subcutaneous tissue, and peritoneal cavity. J. Natl. Cancer Inst. 52:1751- 1755; 1974. Mantyla, M.; Kortekangas, A. E.; Valavaara, R. A.; Nordman, E. M. Tumor regression during radiation treatment as a guide to prognosis. Br. J. Radiol. 52:972-977; 1979. Martial, V. A.; Bosch, A. Radiation induced tumor regression in carcinoma of the uterine cervix: Prognostic significance. Am. J. Roentgenol. 108:113-123; 1970. Ohyama, H.; Yamada, T.; Watanabe, I. Cell volume reduction associated with interphase death in rat thymocytes. Radiat. Res. 85:333-339; 1981. Sobel, S.; Rubin P.; Keller, B.; Poulter, C. Tumor persistence as a predictor of outcome after radiation therapy of head and neck cancers. Int. J. Radiat. Oncol. Biol. Phys. 1:873-880; 1976. Stmad, V.; Tacev, T.; Prokes, B.; Ott, 0.; Krystof, V.; Svobodava, M.; Rasovska, 0.; Konecny, M. Tumor response and treatment complications in radiotherapy of localized prostate cancer. Strahlenther. Onkol. 166; 728-732; 1990. Suit, H. D.; Walker, A. M. Assessment of the response of tumors to radiation: Clinical and experimental studies. Br. J. Cancer 4l(suppl. IV):1 - 10; 1980.
Oncology
Biol.
Phys.. Vol. 34. No. 2, pp. 389-39.X 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved 0360.3016/96 $1.5.00 + .OO
ELSEVIER
0360-3016(95)00222-7
l
Biology Original Contribution IMPACT
OF BIOLOGICAL
KIYOSHI
CLEARANCE
ON TUMOR
RADIORESPONSIVENESS
OHARA, M.D.,*,+ YOSHINORI HAYAKAWA, PH.D.,* HIROSHI HIDEO TATSUZAKI, M.D.*,+ AND YUJI ITAI, M.D.**+
FUJI, M.D.,*
*Department of Radiology, University Hospital, ‘Institute of Clinical Medicine and *Institute of Basic Medical Sciences, University of Tsukuba, Tsukuba City, Japan Purpose: To determine the capacity of biological clearance in tumor regression following radiotherapy by using metastatic brain tumors as a clinical model in which mechanical clearance is negligible. Methods and Materials: Thirty-eight tumors (19 nonsmall cell lung cancer, 11 small cell lung cancer, and 8 nonlung cancer) in 23 patients were followed with computed tomography (CT) scans over 3 months or more following initiation of radiotherapy, with doses ranging between 34 and 66 Gy. The tumor regression rate (RR; mm3/day), which represented the capacity of biological clearance, was calculated for each CT observation period. The complete response (CR) rate was calculated. The relationship between RR and tumor diameter was determined with regression analysis in conjunction with the pattern of contrast enbancement and the type of primary disease. The change of the RR also was examined. Results: The CR rate was 60.5% for the total group; it was lower for ring-enhanced tumors (41.7%) than wy enhanced tumors (69.2%), which included mostly small cell lung cancer metastases. The RR correlated significantly with the tumor diameter (D), with a regression curve of exponential function (RR = 0.035 *D’.§). Tbe RR varied widely and was rather large until 40 days following initiation of radiotherapy, especially for the subgroups of diffusely enhanced tumors and the small cell lung cancer tumors, and became rather constant thereafter. Conclusion: A tumor diameter exponent in the regression curve of smaller than 3.0 indicates that the larger the tumor volume is, the smaller the capacity of biological clearance. The capacity of biological clearance also is dependent on vascularity and cellularity of the tumor components expressed by the pattern of contrast enhancement. Tumor regression, Radioresponsiveness,
Biological clearance, Metastatic brain tumors.
INTRODUCTION
a tumor mass following treatment does not necessarily indicate the presence of viable tumor cells. Instead, the rate of tumor regression following sterilization of the entire clonogenic tumor cell population is determined primarily by the capacity of clearance of these oncologically inactivated components. This involves two main mechanisms; that is, mechanical ckarunce, such as exfoliation and shedding, and biological clearance, such as phagocytic scavenging and liquid permeation. Both
The radioresponsiveness of a tumor has been demonstrated to be of clinical use as a convenient and potent predictor of local control in radiotherapy, especially for head and neck tumors (1,4, 15) and uterine cervical cancer (11, 12, 16). The complete disappearance of tumor following radiotherapy is highly suggestive of a successful outcome, although it is controversial whether rapid initial regression is a prerequisite for this (2, 18). However, complete response is not a perfect predictor because of its rather high false positivity (20); the presence of persistent tumor, immediately following completion of radiotherapy, does not necessarily predict recurrence. Patients with a false positive persistent tumor may be needlessly managed with needless supplementary treatments that are often disabling. In other words, the presence of
mechanisms are involved in tumors on an outer tissue surface, such as head and neck tumors and uterine cervical cancer. However,
only biological
clearance can serve in
tumors completely surrounded by normal tissues that block mechanical clearance. An example of this is a metastatic brain tumor, which has the advantage of being ame-
nable to the assessment of size and texture via computed tomography
Reprints requests to: Kiyoshi Ohara, M.D., Institute of Clinical Medicine, University of Tsukuba, Tsukuba City 305, Japan.
(CT). The purpose of this study is to deter-
Accepted for publication 12 May 1995. 389
390
I. J. Radiation
Oncology
0 Biology
0 Physics
mine the impact of biological clearance on tumor regression using metastatic brain tumors as a clinical model.
METHODS
AND MATERIALS
Sequential CT scans of 23 patients with one or more metastatic brain tumor(s) treated by radiotherapy were evaluated. These CT scansmet the following criteria: (a) a follow-up period of 3 months or more, and a scanning frequency of twice or more following the initiation of radiotherapy; (b) use of contrast enhancement; and (c) tumors treated with a total prescribed dose of at least 30 Gy. The CT scans were obtained with several different types of CT scannersduring a period between 1983 and 1993, but all were obtained with 1 cm slice thickness. The CT follow-up periods ranged from 94 days to 7 years. Eight patients were women and 15 were men, with agesranging between 34 and 88 years. Five patients had a single metastasisand 18 patients had multiple metastases, among which the one or two largest measurabletumors were chosen for the study. Thus, 38 tumors were included: 19 tumors in 12 nonsmall cell lung cancer (NSCLC) patients, 11 tumors in 6 small cell lung cancer (SCLC) patients, 6 tumors in 4 breast cancer patients, and 2 tumors in 1 colon cancer patient. The tumors were treated either with “Co gamma-rays or with 6 MV x-rays, with whole brain irradiation, focal brain irradiation, or both, basically with a weekly dose of 9.0-10.0 Gy/5 fractions. The total prescribed doses ranged between 34.3 Gy/20 fractions and 65.8 Gyl38 fractions: doseslessthan 40.0 Gy for 6 tumors, between 40.0 and 50.0 Gy for 10 tumors, and greater than 50.0 Gy for 22 tumors. In most cases,corticosteroids and/or osmotic diuretics were administered prior to and during the course of radiotherapy. Ommaya’s reservoir was surgically placed for five tumors, three prior to and two during the course of radiotherapy, to reduce tumor masseffect by draining out part of the liquid component of the tumor. Most patients underwent chemotherapy, with a variety of cytotoxic agentsand/or hormone therapy, concurrently or following brain irradiation, for the managementof extracranial disease(s). Tumors were qualitatively classified into two subgroups, basedon the pattern of contrast enhancement: (a) ring enhanced,with central low density; and (b) diffusely enhanced, containing small amounts of low density. The presence of peritumoral edema also was assessed.The volume of each tumor was estimated as a sphere with diameter calculated as the average of the maximum and the minimum diameter on a tumor-center cross-section using two dimensional planimetry; most of the tumors were actually spherical. The complete response(CR) rate was calculated using the Kaplan-Meier method. The tumor regression rate (RR), a daily tumor volume decrement (r&/day), was calculated for each observation interval, and group data was computed based on initial tumor diameter, pattern of contrast enhancement, and the
Volume
34,
Number 2, 1996
type of primary disease (NSCLC, SCLC, and nonlung cancer (non-LC)). The change of RR in conjunction with the time elapsed following initiation of radiotherapy also was examined.
RESULTS Twelve tumors (31.6%) were ring-enhanced and 26 tumors (68.4%) were diffusely enhanced. The pattern of contrast enhancement did not change during the observation period, but calcifications appeared in three tumors. The percentageof ring-enhanced tumors in each subgroup (NSCLC, SCLC, and non-LC) was 36.8% (7 out of 19), 18.2% (2 out of 1l), and 37.5% (3 out of S), respectively. Peritumoral edema was observed in 28 tumors (73.7%) prior to radiotherapy. The tumor volume prior to radiotherapy ranged between 113 mm3 and 83,325 rnrr?, and that of five reservoir-placed tumors ranged between 17,149 mm3 and 83,325 &. The mean volume of the ring-enhanced tumors was significantly larger than that of the diffusely enhanced (16,559.2 mn? vs. 3,623.6 mms, p = 0.0017). All tumors decreasedfollowing radiotherapy with five tumors showing a transient increase (Fig. 1). Twentythree tumors (60.5%), including three calcified tumors, attained CR without any peritumoral edema between 28 days and 27.5 months following initiation of radiotherapy: 5 tumors (41.7%) were ring-enhanced and 18 (69.2%) were diffusely enhanced. The cumulative CR rates at 60, 90, and 120 days following initiation of radiotherapy were 18.7%, 30.0%, and 49.3%, respectively. Those for the diffusely enhanced were 24.0%, 37.3%, and 76.1%, respectively. Fifteen tumors failed to attain CR. Nine of these, however, converted to low density, non-enhancing lesions without peritumoral edema, while the remaining six continued to enhance with contrast. Five tumors recurred by the period between 4 and 10 months following
-30
0
30
ho
w
120 IS0
I x0
Days after RT initiation
Fig. 1. Change of the tumor volume following radiotherapy. The open circles indicate the ring-enhanced tumor and the closed circles the diffusely enhanced tumor. The arrow indicates the placement of Ommaya’s reservoir.
Biological
3.5
. I. *. Y = 2.49X I.-t6
- -,
*.
tumor
clearance
0 K. OHARA
391
et al.
‘.
3
00
.6 .7
.x .9
I
1.1 1.2 I.3 I.4 I.5 1.6 1.7
0
log(x) of tumordiameter(mm)
-
20
.
30
8
60
.
80
Days after
.
’
.
’
4
100 120 l-10 160 180 KT initiation
Fig. 2. Relationshipbetween the tumor regressionrate (RR: mm?/day)and the tumor diameterof entire data. The opencirclesindicatethe ring-enhancedtumor andthe closedcirclesthe diffusely enhancedtumor. The line indicates the regression curve with all tumors.
Fig. 3. Relationshipbetweenthe ratio of tumor regressionrate (RR (observed/expected))andthe time elapsedfollowing initiation of radiotherapy (entire data). The open circles indicate the ring-enhancedtumor and the closedcircles the diffusely enhancedtumor.
treatment; four showed contrast enhancement,
RR (expected) calculated with Eq. 1 and the time elapsed following initiation of radiotherapy was analyzed. The midpoint of each observation period was used as the elapsed time. Those RR data of zero or negative value were included for this analysis. The RR (O/E) varied widely and was rather large until approximately 40 days following initiation of radiotherapy, especially for the diffusely enhanced, and then became nearly constant, showing a biphasic distribution (Fig. 3). There was no significant correlation between them. Therefore, we divided the data into two subgroups based on elapsed time (140 days and >40 days), and again studied the relation between the RR and the tumor diameter. There was a significant correlation between the RR and the tumor diameter (Table 1); however, no significant correlation was observed between the RR (O/E) and the elapsed time.
and only
one had previously attained CR. The following data were excluded from calculation of RR: (a) zero or negative RR values; (b) interval between consecutive scans longer than three months; (c) residual calcification only; and (d) affected by drainage. The RR significantly correlated with tumor diameter (mm) when plotted on a dual logarithmic scale (Fig. 2). With regression analysis the regression curve of the RR was derived as an exponential function of the tumor diameter: A = 0.0347*D2.49,
(Eq. 1)
where A is the RR and D is tumor diameter. An exponential function is normally used in allometry: measuring the growth of a part in relation to a whole living organism. The RR for each subgroup significantly correlated likewise, with the constant and the exponent of the equation shown in Table 1. To examine the chronologic change of the RR following initiation of radiotherapy, a relationship between the RR (O/E); that is, the ratio of the RR (observed) to the
DISCUSSION Radioresponsiveness is an expression not only of radiosensitivity, or a decrease of the tumor cell population, but of a total balance of the decrease and increase in all tumor
Table 1. Constantand exponentof regressioncurve in relationshipbetweentumor regressionrate and tumor diameter Subgroup
Data no.
Constant(c)
Exponent (E)
R
p-Value
Entire Ring-enhanced Diffusely-enhanced NSCLC SCLC Non-LC 5 40 days > 40 days
66 24 42 36 18 12 35 31
0.0347 0.0123 0.0316 0.0407 0.0347 0.0199 0.0276 0.0598
2.49 2.77 2.58 2.41 2.55 2.61 2.63 2.21
0.88 0.84 0.92 0.87 0.85 0.85 0.89 0.87
0.0001 0.0001 0.0001 0.0001 0.0001 0.0004 0.0001 0.0001
The regressioncurve: A = C*D’, whereA is the tumor regressionrate (mrn’/day) in an observationinterval and D is the initial tumor diameter(mm’) in this interval; R: a correlationcoefficient. NSCLC: nonsmallcell lung cancer,SCLC: smallcell lung cancer, non-LC: nonlungcancer, 5 40 days: observationpoint at < 40 days, and > 40 days: at > 40 days.
392
I. J. Radiation
Oncology
0 Biology
l Physics
components. These include (a) cellular components: tumor cells that are proliferative, quiescent, or sterilized, stromal cells, infiltrating cells such as inflammatory cells and macrophages; and (b) organic stromal components: tumor-produced stromal tissues, radiation-induced fibrous tissues, and granulation tissues; and (c) inorganic components: tumor products such as keratin and mucin, necrotic tissues coagulated or liquified, inflammatory and edematous exudates, and so forth. The distribution of these components in tumors will vary with many factors including tumor size, histologic type, and tumor environment (especially vasculature of the tumor and the surrounding normal tissues). In exophytic tumors facing outward, some of these components can be removed by both mechanical and biological clearance; therefore, their individual capacities can not be measured separately. However, in tumors isolated from the outside, all tumor components are removed by biological clearance alone. Accordingly, its capacity can be measured as the RR. If the capacity of biological clearance is proportionate to tumor volume (D’), and tumors are qualitatively identical, the slope of tumor volume regression curves should be identical to each other independent of the tumor volume prior to radiotherapy (5). The slope of the curves obtained, however, was steeper for small tumors than for large tumors. This was shown more clearly by the fact that the RR significantly correlated with the tumor diameter with an exponent of 2.49 of the regression curve. Although the value of the exponent and the constant may be influenced by one another, the exponents of all subgroups were less than 3.0. This indicates that the smaller the tumor volume is, the larger the RR. Moreover, it was as small as 2.21 for the subgroup of 240 days, suggesting that the RR is proportionate to the tumor surface area (0’). This would suggest that biological clearance arises mainly at the peripheral zone of the tumor; that is, the tumor-normal tissue interface. The RR (O/E) values varied widely and were rather large in the initial phase of 40 days following radiotherapy, particularly for the diffusely enhanced tumors. Actually, the diffusely enhanced tumors attained CR more rapidly and to a higher rate than the ring-enhanced tumors. The higher initial RR (O/E) value of the diffusely enhanced tumors is due partly to the fact that this group included most of the data from the subgroup of SCLC, a tumor known to be more radiosensitive than the others. This higher initial rate has been observed also by Kataoka et al., although the type of primary disease was not described (13). The RR (O/E), however, did not correlate with the elapsed time following initiation of radiotherapy. This lack of correlation may be due to the short and irregular observation period of up to 4 months and the small number of patients. However, the fact that a tumor mass of low density without contrast enhancement persisted for 8 months could support the biphasic response in tumor regression. Bataini et al. have reported the biphasic distribution in
Volume
34, Number
2, 1996
tumor regression (3). The change of the CR rate of metastatic cervical lymph nodes from primary disease of the oropharynx and pharyngolarynx showed an initial steep linear phase followed by a period of constant decrease in the slope. The CR rate was significantly lower for lymph nodes than for the primary disease at the end of radiotherapy, but the percentage of cases for each group with residual disease were almost equal 2 months later. This biphasic response was also observed for some salivary gland tumors, which were treated with neutrons (5). The pattern of tumor regression of prostate cancer reported by Stmad et al. strongly suggests a biphasic response as well (19). While these tumors were all isolated from outer surfaces, no qualitative assessment was made. One hypothesis for the biphasic response, proposed by Bataini et al. suggests that radiation-induced impairment of blood flow reduces the clearance rate (3). This hypothesis is supported by the fact that tumor regression is prolonged by radiation-induced stromal cell damage that was shown experimentally with a model of the severe combined immunodeficient mouse (6). Bessel1 et al. have explained the cause for the biphasic response as follows: The regression depends on the ratio of proliferating cells to stroma in which the initial regression represents the rapid clearance of the dead cellular component, followed by the slow clearance of the stromal component (5). In metastatic brain tumors, the area of central low density is considered to be composed of biologically inactivated components, mostly of liquified necrosis and possibly exudates, because they were readily drained through a reservoir in large tumors. In this context, the ring-enhanced tumors contain rather small amounts of vital components of tumor cells and tumor stromal tissues, which exist only at the peripheral rim of the tumor mass. Conversely, the diffusely enhanced tumors consist largely of vital components distributed throughout the tumor mass. These enhancement patterns suggest alternate explanations for the biphasic response. The first is that the tumors in the initial phase contain a proportionately larger amount of vital cellular components, especially in the diffusely enhanced tumors, which decrease volume by cytolysis and demonstrate cell volume reduction with interphase cell death or apoptosis (8,17) induced by irradiation. The second is that the tumor cells that are well nourished by the functional stroma should be more radiosensitive. The third is that biologic clearance can arise, even deep within a diffusely enhanced tumor, through the functional tumor stroma. The capacity of biological clearance can differ with the type and conditions of normal tissues in which the tumor has arisen. In experimental studies comparing the rate of dead tumor cell removal from brain, muscle, and SC tissues of rats, it is most rapid in muscle and slowest in brain (14). This difference may be due to qualitative and quantitative differences in these tissues including the stroma and scavenging cells, which are microglia and astrocyte in the brain (10). Moreover, the rate of biological clearance in these
Biological
tumor
clearance
normal tissues can be affected by irradiation, probably in a dose-dependent fashion. Cairncross et al. have reported histopathologic findings of me&static brain tumors that persisted for months following radiotherapy (7). The tumor mass lesions showed discrete, well-circumscribed regions of necrosis surrounded by gliosis without tumor cells. Our nine tumors, which persisted as low-density masses without contrast enhancement or petitumoral edema, may have been this type of sterilized tumor. Magnetic resonance imaging (MRI), which provides more information about tumor quality, has demonstrated large endophytic uterine cervical tumors that persisted at 6 months following radiotherapy and resolved at 9 months (9). These tumors with slow regression showed declining signal intensity of the solid component, which probably represents inor-
0 K. OHARA
et al.
393
ganic elements such as coagulated necrosis and fibrous tissues. The solid component may be more slowly removed than the liquid component, because permeation clearance can also serve for the latter. Thus, the tumor quality following radiotherapy is important to predict local tumor control. In conclusion, evaluation of the RR of metastatic brain tumors following radiotherapy reveals clearance as a function of an intermediate range between the tumor volume (D’) and the tumor surface area (D’). Therefore, the larger the tumor volume, the smaller the capacity of biological clearance. In addition, this is larger in tumors with less necrosis than in tumors with a large amount of necrosis, particularly in the early phase following initiation of radiotherapy.
REFERENCES 1. Barkley, H. T., Jr.; Fletcher, G. H. The significance of residual disease after external irradiation of squamous-cell carcinoma of the oropharynx. Radiology 124:493-495; 1977. 2. Bartelink, H. Prognostic value of the regression rate of neck node metastases during radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 9:993-996; 1983. 3. Bataini, J. P.; Bemier, J.; Jaulerry, C.; Brunin, F.; Pontvert, D.; Lave, C. Impact of neck node radioresponsiveness on the regional control probability in patients with oropharynx and pharyngolarynx cancers managed by definitive radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 13:817-824; 1987. 4. Bataini, J. P.; Jaulerry, C.; Brunin, F.; Ponvert, D.; Ghossein, N. A. Significance and therapeutic implications of tumor regression following radiotherapy in patients treated for squamous cell carcinoma of the oropharynx and pharyngolarynx. Head Neck 12:41-49; 1990. 5. Bessel, E. M.; Catterall, M. The regression of tumors of the head and neck treated with neutrons. Int. J. Radiat. Oncol. Biol. Phys. 9:799-807; 1983. 6. Budach, W.; Taghian, A.; Freeman, J.; Gioioso, D.; Suit, H. D. Impact of stromal sensitivity on radiation response of tumors. J. Natl. Cancer Inst. 85:988-993; 1993. 7. Caimcross, J. G.; Chemik, N. L.; Kim, J.; Posner J. B. Sterilization of cerebral metastases by radiation therapy. Neurology 29:1195-1202; 1979. 8. Falkvoll, K. H. The relationship between changes in tumor volume, tumor cell density and parenchymal cord radius in a human melanoma xenograft exposed to single dose irradiation. Acta Oncol. 29:935-939; 1990. 9. Flueckiger, F.; Ebner, F.; Poschauko, H.; Tamussino, K.; Einspieler, R.; Ranner, G. Cervical cancers: Serial MR imaging before and after primary radiation therapy: A 2-year follow-up study. Radiology 184:89-93; 1992. 10. Gueneau, G.; Drouet, J.; Privat, A; Court, L. Differential radiosensitivity of neurons and neuroglia of the hippocam-
11.
12.
13.
14.
15. 16. 17. 18.
19.
20.
pus in the adult rabbit. Acta Neuropathol. 48:199-209; 1979. Hardt, N.; van Nagel, J. R.; Hanson, M.; Donaldson, E.; Yoneda, J.; Maruyama, Y. Radiation-induced tumor regression as a prognostic factor in patients with invasive cervical cancer. Cancer 49:35-39; 1982. Jacobs, A. J.; Faris, C.; Perez, C. A.; Kao, M.; Galakatos, A.; Camel, H. A. Short-term persistence of carcinoma of the uterine cervix after radiation: An indicator of long-term prognosis. Cancer 57:944-950; 1986. Kataoka, M.; Inatsuki, S.; Kawamura, M.; Itoh, H.; Hamamoto, K. Treatment response on CT scan in patients with brain metastases treated with irradiation. Radiat. Med. 9:41-46; 1991. Kumar, R. V.; Hoshino, T.; Wheeler, K. T.; Barker, M.; Wilson, C. B. Comparative rates of dead tumor cell removal from brain, muscle, subcutaneous tissue, and peritoneal cavity. J. Natl. Cancer Inst. 52:1751- 1755; 1974. Mantyla, M.; Kortekangas, A. E.; Valavaara, R. A.; Nordman, E. M. Tumor regression during radiation treatment as a guide to prognosis. Br. J. Radiol. 52:972-977; 1979. Martial, V. A.; Bosch, A. Radiation induced tumor regression in carcinoma of the uterine cervix: Prognostic significance. Am. J. Roentgenol. 108:113-123; 1970. Ohyama, H.; Yamada, T.; Watanabe, I. Cell volume reduction associated with interphase death in rat thymocytes. Radiat. Res. 85:333-339; 1981. Sobel, S.; Rubin P.; Keller, B.; Poulter, C. Tumor persistence as a predictor of outcome after radiation therapy of head and neck cancers. Int. J. Radiat. Oncol. Biol. Phys. 1:873-880; 1976. Stmad, V.; Tacev, T.; Prokes, B.; Ott, 0.; Krystof, V.; Svobodava, M.; Rasovska, 0.; Konecny, M. Tumor response and treatment complications in radiotherapy of localized prostate cancer. Strahlenther. Onkol. 166; 728-732; 1990. Suit, H. D.; Walker, A. M. Assessment of the response of tumors to radiation: Clinical and experimental studies. Br. J. Cancer 4l(suppl. IV):1 - 10; 1980.