688 CLINICAL RADIOLOGY

The groups were comparable with respect to histology, age and stage. The mean follow-up period was 4.3 and 2.6 years, with minor complication rates of 25% (95% confidence limits 13 to 37%) and 16% (95% CL 6 to 26%) in control and study groups, respectively. (This includes one complication in 13 study patients treated with intracavi- tary radiotherapy only.) The only major complication was one second malignancy in the treatment area and there was no difference in actuarial survival. Treatment was out-patient-based, required minimal analgesia, less radioprotection than alternative isotopes and elimin- ated the risk to staff of manual afterloading systems. We conclude that the Buchler system is a safe, convenient substitute for other methods without increased recurrence or complication rates.

PINEAL GERMINOMAS IN CHILDHOOD. T R E A T M E N T CONSIDERATIONS D. JENKIN, H. CHAN, M. GREENBERG, B. HENDRICK, H. HOFFMAN, R. HUMPHREYS, M. SONLEY and S. WEITZMAN The Hospital for Sick Children, Toronto, Canada

From 1959 through 1986, 50 consecutive children were diagnosed as having a pineal region tumour at the Hospital for Sick Children. A tissue diagnosis was available in 35 of 50 (70%). Thirty-one children were treated as having a germinoma, 21 from the biopsy era and 10 from the pre-biopsy era. Four of 31 (13%) patients had tumour at additional local sites (suprasellar two patients, thalamic two patients). Seven of 31 (23%) had diabetes insipidus at diagnosis.

In 27 of 31 (88%) a shunt was required. In 10 of 18 (55%) biopsied patients a partial or subtotal tumour resection was undertaken. There were no operative deaths. Thirty of 31 (97%) patients were irradiated after surgical treatment. The irradiated volume was local, 11; local+brain, 8; and craniospinal in 10 patients. The 10-year survival and relapse-free survival rates were 90% and 86% respectively for the 21 biopsy era patients. Only the one unirradiated patient died due to tumour progression. Spinal relapse occurred in two of 19 (11%) patients who received less than craniospinal irradiation. Both were salvaged from relapse. For the pre-biopsy era 'germinomas' (n=10) survival and relapse-free survival rates at 10 years were 89% and 78% respectively. Two patients suffered late local relapse and chronic progression of local disease to die 8 and 10 years from diagnosis. This form of progression virtually excludes germinoma as the appropriate diagnosis. Overall we have no confirmed example of death due to progressive germinoma among 30 irradiated patients.

It is difficult to justify the overall intensification of treatment that has occurred over the years in this highly curable tumour. New treat- ment proposals will be made.

THE ROLE OF RADIOTHERAPY IN IN-SITU CARCINOMA OF THE LARYNX P. MacLEOD and F. DANIEL Plymouth General Hospital, Plymouth

In the period 1966-1982, 194 patients have been referred to this department for definitive treatment of in-situ (Tis) and invasive squamous carcinomas of the larynx. In-situ lesions represent 10% of this group, (20 patients) stage T1 37%, T2 25% and stages T3 and T4 28% (UICC). We have analysed our results for the treatment of in-situ lesions and present them in comparison to TI, invasive lesions. All patients were treated with megavoltage radiation, Co 60 or 4 MeV photons, using two fields, laterals or anterior obliques, mean size 4.5x4.5 cm. The mean doses given were 1759 ret (Tis lesions) and 1743 r et (T1 lesions).

In the Tis group six of 20 patients (30%) failed to achieve local control with radiotherapy. The surgical augmented salvage rate was 90% (18 of 20), one patient dying an intercurrent death and the other still alive, both with uncontrolled primary disease. Twelve of 71 patients (17%) in the T1 group failed local control with radiotherapy. The augmented salvage rate was 94% (68 of 71). Statistically there is no difference between the primary control and augmented salvage rate in the Tis or T1 groups (X 2 P>0.2). Overall actuarial survival at 5 and 10 years is 100% and 100% respectively in the Tis group and 96.8% and 92.6% in the T1 group (P=0.275). Five and 10 year actuarial recurrence free survival rates are 78.8% and 69% respectively for the Tis group and 86.1% and 78.4% for the T1 group (P=0,548).

In this series we have found that the local control and survival rates are not significantly different for in-situ and stage T1 invasive car- cinomas of the larynx and we support the radiotherapeutic approach to the management of in-situ lesions.

RADIOTHERAPY FOR E P I D E R M O I D ANAL CARCINOMA D. OTIM-OYET, A. HORWICH, J. CROW, H. T. FORD and C. FISHER Royal Marsden Hospital, Sutton, Surrey

A retrospective analysis on 75 patients treated at the Royal Marsden Hospital between 1958 and 1986 showed that 74.6% presented with Lyons T2 (29) and T3 (27) tumours. Twenty patients (26.6%) had inguinal nodal metastases and three systemic dissemination. Defini- tive treatment was radiotherapy in 55 patients, 46 of whom received a radical dose of 60 Gy/30#/6 weeks, or NSD equivalent. Four of these had combined modality treatment with 5FU and mitomycin-C. Of 20 patients treated by surgery, nine received pre- or post-operative radio- therapy. Overall the median follow-up of survivors was 49.8 months. The actuarial cause specific survival at 3 years was 55% and this was adversely influenced by increasing stage and histological grade. In the subgroup receiving radical radiotherapy, 26 of 42 (61.9%) evaluable patients achieved primary tumour control. However for T1 T2 NO tumours this rate was 87.5%. Three of five (60%) relapsing patients were salvaged by surgery. Six of 20 (30%) patients treated by surgery recurred locoregionally and none were salvaged by radiotherapy. Eight of 13 (61.5%) patients developing distant metastases also had uncontrolled primary tumours.

It is concluded that radiotherapy alone is effective treatment for small epidermoid anal carcinomata, and has the advantage of preserv- ing anal continence. More advanced tumours probably require com- bined modality treatment.

THE IMPORTANCE OF SET-UP REPRODUCIBILITY IN CONFORMATION THERAPY A. T. TATE Royal Free Hospital, London

Computer-aided three-dimensional tumour imaging and target recon- struction is now possible and, using computer-controlled treatment machines, complex radiation dose distributions can be achieved. The exact shaping of these distributions around a target can be confirmed using an anthropomorphic phantom. In practise, however, the final accuracy of spatial dose delivery is dependent upon exact and reproducible patient positioning. The greatest clinical advantage of three-dimensional or conformation therapy is seen when the target is small and lies deep within a body cavity. The standard setting up techniques used for large field conventional therapy are not suffi- ciently accurate for these long, small diameter targets.

Simple, practical methods for improved patient immobilisation and set-up are described and measurements which demonstrate the limits of daily reproducibility are presented.

Potential methods of improving the verification of the accurate location of the treatment volume within the patient are discussed.

THE CLINICAL T O L E R A N C E OF COMBINED • RADIOSENSITISERS Ro 03-8799 AND SR 2508

T. S. MAUGHAN, N. M. BLEEHEN, H. F. V. NEWMAN and P. WORKMAN Addenbrooke's Hospital, Cambridge

The hypoxic ceil radiosensitisers Ro 03-8799 (Pimonidazole) and SR 2508 (Etanidazole) have different clinical toxicities, and therefore combining the two drugs may improve radiosensitisation without increasing toxicity. Ro 03-8799 produces a transient central nervous system syndrome which limits individual doses to i g/m 2 while SR 2508 results in a cumulative peripheral neuropathy limiting the total dose to 34 g/m 2 at 2 g/m 2 individual doses. The present study reports the clinical tolerance following multiple doses of 2 g]m 2 SR 2508 and 0.75 g/m 2 Ro 03-8799 given three times a week. Of 27 patients treated, four have received 15 infusions and all four developed WHO Grade II peripheral neuropathy. No peripheral neuropathy was reported fol- lowing this dose of SR 2508 alone, therefore some interaction between the drugs in terms of chronic peripheral neurotoxicity does exist. Pharmacokinetic studies confirm this suggestion, as the most signifi- cant predictor for the development of peripheral neuropathy is the cumulative area under the curve for Ro 03-8799.

The combination at the maximum tolerated dose level produces a single dose sensitiser enhancement ratio of about 1:5. Despite the increased incidence of neurotoxicity compared with SR 2508 alone, an enhanced sensitiser efficacy may be obtained using the two drugs in combination over either drug used alone. This approach may be tested in a randomised setting once the current Phase 3 studies using the two sensitisers individually have been concluded.