TIME TO RECURRENCE OF SQUAMOUS CELL CARCINOMA OF THE HEAD AND NECK Philip M. Stell, ChM, FRCS
A personal series of 3,215 patients with proven squamous cell carcinoma of the mucosal surfaces of the upper aerodigestive tract is presented. Of these patients, 515 suffered a recurrence of their primary tumor after a course of radiotherapy. The time from the end of radiotherapy to diagnosis of a primary recurrence was not related to any known host factors, nor to tumor factors except site. However, it was the most significant predictor of survival, both from initial presentationand from the date of recurrence. The time to recurrence did not relate to the size of the tumor at recurrence, suggesting that large recurrences are multicentric. The time to recurrence also predicted response and survival of 67 patients with end-stage disease treated by chemotherapy. The response and survival were better in tumors with a time to recurrence longer than the median; the increased survival was not significant, but the increased response was. HEAD 81NECK 1991;13:277-281
cently begun to be studied in human head and neck cancer. It is generally assumed that the doubling time of a tumor is exponential, in at least part of its course.' That is, the volume doubling time of the tumor is constant, or, put another way, the volume of the tumor doubles in a given time interval, and then doubles again in the same, succeeding, time interval and so on. Thus, if V, is the initial volume, and V, the volume a h r time t and &r n doublings:
ie, T h e kinetics of tumor growth may be studied at 3 levels: (1)cellular kinetics; (2) the macroscopic growth rate; and (3) the time interval between major events. Cellular kinetics have only re-
VJV0=2" ie, log,(VJV,) =n.log$ ie,
From the Department Otorhinolaryngology, University of Liverpool, Liverpool, England. Acknowledgments: The author is grateful to Mrs. Jill Deeprose and Mrs. Brenda Cowley for the typing, to the North West Cancer Research Fund for financial support. and to Dr. J.E. Dalby of the Mersey Region Radiotherapy and Oncology Centre for advice on radiation dosage. Address reprint requests to Professor Stell at the Department of Otorhinolaryngology, University of Liverpool, Royal Liverpool Hosptal, P.O. Box 147,Liverpool L69 3BX, England. Accepted for publication December 24,1990 CCC 0148-64O3BI1040277-05 $04.00 0 1991 John Wiley & Sons, Inc.
Recurrence Time of Squamous Cell Carcinoma
If the Volume doubling time is d, the time t taken to undergo doublings is . d le,
d =tln Substituting in Eq. (2) above: d= t~log$/(loge(VJV,)
HEAD & NECK
If we know V,,and V ,the volume doubling time can thus be calculated readily. V, is often assumed to be the volume of 1 celL5 Few data are available about the growth rate of human tumors. The reason for this is obvious: it is not often possible in the human to measure the size of a tumor at more than 1 point of its growth. One of the common techniques for assessing tumor growth is the serial measurement of pulmonary metastases. Collins et a1.2 measured the growth rate of pulmonary metastases on serial radiographs of 24 patients. Their series included 2 head and neck tumors (1 from the larynx and 1 from the lip), and these tumors had doubling times of 37 and 131 days, respectively.’ Based on serial measurements of tumor volume, Shackney et al. divided solid tumors into 3 group^.^ The first group contained Ewing’s sarcoma, testicular carcinoma, and the nonHodgkin’s lymphomas and had a mean volume doubling time of less than 30 days. Hodgkin’s disease, osteogenic sarcoma, and fibrosarcoma were in an intermediate group with a mean volume doubling time ranging from 30 to 70 days. The third group was formed by adenocarcinoma and squamous cell carcinoma of the lung and adenocarcinoma of the colon, with a mean doubling time exceeding 70 days. The authors stated that rapidly growing tumors in the first group are those most often responsive to chemotherapy, and these responses are often durable. However the situation is not quite this simple, because later it was found that Hodgkin’s lymphoma, placed in the intermediate group (also containing osteogenic and fibrosarcoma) of tumors resistant to chemotherapy, was most successfully treated by chemotherapy. Data about squamous cell carcinoma of the head and neck in the human are sparse indeed. Galante and colleagues4 measured the doubling time of 17 local and recurrent head and neck tumors. They used the technique for measuring the progress of recurrent nodules of breast cancer previously described by Philippe and Le Gal,5 which assumes that all recurrences arise from a single cell. They found a median value of 9.5 days. They also abstracted data from the literature on 31 other head and neck tumors and found the doubling times of pulmonary metastases of head and neck tumors to vary between 15 and 320 days. They claimed that local recurrences grow faster than pulmonary me-
Recurrence Time of Squamous Cell Carcinoma
tastases, and that site and age do not influence growth rate, but sex does. Steel abstracted the volume doubling time of 27 squamous cell tumors of the head and neck from the literature and found a median of 57 days with 95% confidence interval of 47 to 75 days. This result was based on actual measurement at 2 or more points in time and not on an assumption that tumors repopulate from 1 cell, and must be accepted as the “gold standard.” It is, of course, subject to several errors as listed by Steel.‘ The time to recurrence is easy to measure and presumably reflects growth rate, but does not appear to have been studied, and thus deserves study in relation to response t o treatment, host, and tumor factors, and to prognosis. MATERIALS AND METHODS
This study is based on a personal series of 3,215 patients with squamous cell carcinoma of the head and neck, managed in a 27-year period from 1963 to 1990. The data of all patients were recorded at the time the patient was first seen, initially on cards, and after 1977 on a microprocessor. The data were collected prospectively: the data of the follow-up is known from personal observation, the patient’s general practitioner, the Mersey Regional Cancer Registry, or the National Health Service Registry. Five hundred fifteen patients treated by irradiation for a histologically proven squamous cell carcinoma of the mucosal surfaces of the upper respiratory tract, and who later suffered a recurrence at the primary site, were chosen from this data bank. Two hundred thirty-seven patients were previously untreated, and 278 were referred for treatment of recurrence. One (0.2%)of these patients has been lost to follow-up. The median potential period of follow-up is 11 years. From 1963 to 1973, previously untreated patients were treated by 5,250-5,500 rads in 3 weeks, and from 1974 onward by 6,250-6,500 rads in 5 weeks. The tumors were classified by the latest UICC system7 and their general condition by the ECOG performance status scale.’ The data on performance status were insufficient for classification of 16 patients seen in earlier years, and the data on T and N stage were insufficient for 17 patients to allow staging by the latest criteria. The tumor was given the appropriate rT stage when it recurred. Patients with a recurrence presenting more
HEAD & NECK
than 8 years after initial treatment were excluded, and their tumor was counted as a second metachronous tumor, even if the recurrence was in exactly the same place as the first tumor. The reason for this decision is that the 95% upper confidence limit of the volume doubling time of squamous cell carcinoma of the head and neck is 75 days6 Thus, a tumor arising from 1 cell and growing at the slowest rate will reach a diameter of 1 cm (and thus be clinically detectable) about 8 years later. The time t o recurrence of the primary tumor was calculated from the end of radiotherapy. Patients with a tumor that was still clinically visable 3 months after the end of radiotherapy, and from which a positive biopsy was obtained, were counted as having persistent disease and were excluded. The data for host and tumor factors relative to time to primary recurrence are shown in Tables 1 and 2. The relation between time to recurrence and the various host and tumor factors was analyzed by multilinear regression using a logarithmic transformation, because the data for recurrence times are right-skewed. The effect of time to recurrence on overall survival and survival after recurrence was calculated using univariate methods? The factors identified by univariate methods were then subjected to Cox's multivariate analysis" to exclude interdependence between factors. Qualitative data are shown in contingency tables and analyzed by x2. Confidence intervals are given where appropriate." The time to recurrence was very skewed to the right, so that median values are quoted.
Table 1. Time to recurrence (weeks): host factors.
Age 60 Sex Men Women Performance status 0 I-IV Previous treatment elsewhere
33 (28-41) 36 (28-41)
34 (28-40) 36 (27-41)
157 64 278
33 (27-38) 36 (25-46) 36 (28-41)
Patients treated after 1973 with a higher dose in 5 weeks had a slightly shorter median time to recurrence of 343 days compared to 382 days for patients treated with a lower dose in 3 weeks. However, multilinear regression showed this difference to be entirely nonsignificant (t235 = 0.05).
Overall Survival. Time to recurrence had a dramatic impact on survival time. Figure 1 shows that the overall survival for patients with a time to primary recurrence above the median was about 20%better at 5 years than those below the median, and this difference was highly significant (x21 = 72.3, p < 0.001). Cox's multivariate analysis (Table 3 shows that time t o recurrence was the most highly significant predictor of survival, when relations be-
Table 2. Time to recurrence (weeks): tumor factors. n
Time to Primary Recurrence.
Recurrence Time of Squamous Cell Carcinoma
Median (95% CI)
Cl, confidence interval.
The time (median and range) to primary recurrence was 36 weeks (10- 384 weeks). The data for time to recurrence for host and tumor factors are shown in Tables 1 and 2. It can be readily seen that time to recurrence does not vary much with the various host factors, and multilinear regression showed that none of these differences were significant. The time to recurrence varied with site, from 27 weeks for tumors of the mouth to 41 weeks for tumors of the larynx, and this difference was highly significant (t235 = 3.64, p < 0.001).Although the time to recurrence of well-differentiated tumors was shorter than that of other histologic grades, this difference was not significant (t23= 1.7).
Site Mouth Larynx Hypopharynx Oropharynx Nasopharynxlear Nose & sinuses Histologic grade Well-differentiated Moderately differentiated Poorly differentiated Ungraded Stage I
Median (95% CI)
111 202 85 57 34 26
24 (19-28) 41 (36-46) 27 (22-36) 32 (23-41) 40 (22-109) 37 (25-45)
136 121 102 156
31 (27-37) 36 (26-41) 36 (28-45) 36 (28-45)
73 62 71 31
39 (25-54) 30 (23-37) 40 (29-46) 27 (21-31)
HEAD & NECK
x-A A-A A-A
FIGURE 1. Survival related to time to recurrence: (A) patients with time to recurrence greater than the median; (0)patients with time to recurrence less than the median.
tween interdependent factors are taken into account. Survival After Primary R.cuwence. The time to recurrence affected the survival rate aRer treatment of such a recurrence. The survival, 5 years after a primary recurrence, varied from 19% for those tumors that recurred within 1year, to 48% for those tumors that recurred more than 4 years after initial treatment (Table 4). This effect was highly significant (x22 = 21.2,p c 0.001).
The time to recurrence was longer for patients with ffl tumors than those with ff4 tumors (Table 51, but this difference was not significant. Time to Recurrence Related to T Stage.
Sixty-seven patients, who developed recurrent primary tumors that were not suitable for further radical treatment, were treated by chemotherapy. Seventeen of these 67 patients had a response (partial or complete), and 15 of these had a time to recurrence greater than the median. This difference was Chemotherapy.
Table 3. Predictors of survival.
Time to recurrence Site General condition Stage Age Histology Sex
4.21 3.49 2.23 2.03 1.55 0.70 0.37