Continuous carboplatin infusion during 6 weeks' radiotherapy in locally inoperable non-small-cell lung cancer: a phase I and pharmacokinetic study.

A phase I study was performed in 21 patients with previously untreated, locally inoperable, non-small-cell lung cancer (NSCLC) with ambulatory continuous carboplatin infusion together with continuous thoracic irradiation over 6 weeks. A dose range for carboplatin of 15 mg m-2 day-1 during the last 21 days (first level), during the last 31 days (second level), or during 6 weeks of the radiation period (third level) and thereafter 20 or 25 mg m-2 day-1 during 6 weeks of radiation (fourth and fifth level) was used. The total radiation dose was 60 Gy given as 2 Gy day-1 for 5 days week-1. The first three patients received radiotherapy without carboplatin. WHO grade III/IV leucopenia and thrombocytopenia occurred in the last two dose levels in two out of six and one out of six patients with 20 mg m-2 day-1 respectively, and in all three patients with 25 mg m-2 day-1 (dose-limiting toxicity). One local infection around the port and a subclavian vein thrombosis occurred. Radiation toxicity of the lung and oesophagus did not seem to be influenced by carboplatin treatment. Out of 21 patients one had a complete response (CR), ten partial response (PR), six stable disease (SD) and four progressive disease (PD). Total (TPt) and ultrafilterable plasma platinum (UPt) were measured in the last three dose levels with atomic absorption spectrophotometry with Zeeman correction. The mean (s.d.) level for TPt for 6 weeks at 15, 20 and 25 mg m-2 day-1 was 0.76 (0.15), 0.78 (0.19) and 0.90 (0.22) mg l-1 for UPt 0.10 (0.03), 0.12 (0.02) and 0.20 (0.03) mg l-1 respectively. TPt concentration levelled off after 3 weeks. The mean (s.d.) CLTB for UPt was 281 +/- 21 ml min-1 and correlated with glomerular filtration rate (r = 0.61, P = 0.03). As estimated with the sigmoid Emax model defined by the Hill equation the percentage reduction in platelets correlated with the area under the curve for UPt (r = 0.77). The maximum tolerable dose of carboplatin with concomitant continuous 60 Gy radiotherapy is 25 mg m-2 day-1; the recommended dose for phase II or III studies is 20 mg m-2 day-1 day for 6 weeks.

The standard treatment for stage III non-small-cell lung cancer (NSCLC) is local thoracic irradiation. In 1973 a Radiation Therapy Oncology Group trial showed a significant relationship between higher total dose of radiation and control of tumour within the field as well as survival at 2 years (Perez et al., 1982). However, the high rate of distant metastases and the still frequent failure within the irradiated volume made the improvement in prognosis rather limited (Perez et al., 1987). These failures provided a rationale for the combination of radiation and chemotherapy.
For several years induction chemotherapy followed by radiotherapy was used to reduce local and distant recurrences and to improve survival. Some randomised trials showed improved survival with this approach (Dillman et al., 1990;LeChevalier et al., 1992), however others could not demonstrate the value of induction chemotherapy (Cullen et al., 1991;Mattson et al., 1991;Morton et al., 1991). A new approach would be simultaneous administration of chemotherapy and radiation therapy. In addition to a cytoreduction by both treatments, a radiosensitising effect might be of extra benefit. In vitro, radiosensitising effects have been found for cisplatin and carboplatin (Douple, 1979;Douple et al., 1985;Begg et al., 1987;Skov and MacPhail, 1991). Furthermore, platinum-containing regimens in adequate doses have shown objective responses to clinical trials with NSCLC (Bunn, 1989). In at least one out of four randomised trials concurrent chemoradiation showed improved local control and doubling of the 2 year survival at the cost of considerable toxicity (Soresi et al., 1988;Ansari et al., 1991;Trovo et al., 1991;Schaake-Koning et al., 1992). Car-boplatin has reduced most of the toxicities of cisplatin treatment, especially when it is administered as a continuous infusion (Smit et al., 1991). In the present study the feasibility and optimal dose of continuous carboplatin infusion on an outpatient basis during local thoracic irradiation in stage III NSCLC was analysed.

Patients and methods
Patients Patients with histologically proven NSCLC were eligible if they fulfilled the following criteria: age below 75 years, locally unresectable NSCLC without supraclavicular nodes, Eastern Cooperative Oncology Group (ECOG) performance score < 2, serum creatinine < 120 tmol ' or creatinine clearance > 60 ml minl, serum bilirubin < 2.0 mg dl-' leucocytes > 3.0 x I0 1'-and platelets > 100 x I0 I-1. Patients with prior chemoor radiotherapy were ineligible. Written informed consent was obtained from all patients. The study was approved by the Medical Ethical Committee of the University Hospital of Groningen.

Carboplatin
Carboplatin was supplied by Bristol Myers Squibb and dissolved in 5% glucose. Every 48 h patients were provided with 15 ml of freshly constituted solution in a 20 ml syringe. A portable battery-powered syringe driver (Graseby Medical MS 16A) was connected to a venous access port (Infuse-A-Port) with an extension tube and a Huber needle. The venous access was attained by a standard subclavian vein puncture. The subclavian line was a silicone rubber catheter, tunnelled under local anaesthesia to a subcutaneously implanted metal injection port (VAP) (Greidanus et al., 1987). Patients were carefully instructed by an oncology nurse how to change syringes in the portable pump. The VAP was checked for local infection every week; the needle and extension tube were renewed in the third week of treatment.

Irradiation
Continuous thoracic irradiation delivered by a linear accelerator with megavoltage photons (6 MV) was applied for 5 days a week for 6 weeks. The total dose was 60 Gy in 30 fractions of 2 Gy. Simulation was performed in all patients. The target volume encompassed all visible local and regional disease with a 2 cm margin of normal tissue that is based on examination of the CT scan of the thorax before the start of the treatment. The target volume until 40 Gy included also the mediastinum from 2 cm on above the suprasternal notch to 5 cm below the carina and extended to 2 cm across the midline.
Until 40 Gy anterior-posterior opposed fields were both treated at each treatment session. Thereafter, irradiation was given to the original tumour site with a 2 cm margin using a three-field technique or oblique fields to exclude the spinal cord. No corrections for pulmonary field irradiation were made. The dose compliance was within 5% of the planned dose after recalculations from the daily dose records.

Study design
Three patients received radiotherapy only. Thereafter, the first dose level of carboplatin was 15 mg m2 day-' given during the last 21 days of the radiotherapy. The next level was given during the last 31 days and the following level during 6 weeks of radiotherapy. Further escalations in dose were planned at 5 mg m2 day-' during 6 weeks of continuous irradiation.
Unacceptable toxicity was defined as non-haematological toxicity (except alopecia) exceeding (but not including) WHO grade II and/or haematological toxicity exceeding grade III. Acute toxicity was defined as toxicity that appeared from the start of treatment until 2 months after the end of the combined treatment. At least three patients were enrolled in each dose level. If unacceptable toxicity occurs in one out of three patients then a total of six patients will be enrolled in that level.
Patients were assessed with complete blood counts, liver and renal function, weight, performance score and nonhaematological (WHO) toxicity score. Moreover, acute pulmonary toxicity was scored with chest radiographs every 2 weeks, lung function tests with diffusion capacity measurements (according to the recommendations by the American Thoracic Society) including pulmonary membrane function and capillary blood volume before, midway and 2 weeks after finishing treatment. A CT scan of the chest was performed before and 3 weeks after the end of treatment. Other non-haematological toxicity was scored independently by two investigators using WHO criteria. The highest score was used. Tumour response was measured according to WHO criteria 3 weeks after the end of the combined treatment (WHO, 1978). Total (TPt) and ultrafilterable plasma platinum (UPt) were measured with atomic absorption spectrophotometry with Zeeman background correction. Samples were collected weekly over 6 weeks of treatment in heparinised tubes (Venoject; Omnilab, Breda, The Netherlands) on ice. For Tpt, samples were centrifuged (1500 g for 5 min at ambient temperature). An aliquot of 2 ml of plasma was then directly ultracentrifuged (1800 g for 60 min at ambient temperature) with an Amicon Centifree micropartition system with YHT membranes (Amicon, Oosterhout, The Netherlands) to make UPt samples. Samples were stored at -20°C to await analysis. The lower detection limit for TPt and UPt is 0.1 mg ml-' (s.d. 7.7%) and 0.025mg 1' (s.d. 5%) respectively. Areas under the curve (AUC) for TPt and UPt were calculated by the trapezoidal rule.
The pharmacodynamic linkage between the percentage reduction in platelets and AUC to ultrafilterable platinum has been modelled with the modified Hill equation (Egorin et al., 1994): where E represents the percentage decrease in platelets produced by AUC, which is the area under the concentration curve for ultrafilterable platinum, Emax represents the maximum elicitable effect of 100% decrease in platelets and AUC" represents the AUC associated with 50% of Emu. H is the Hill constant describing the sigmoidicity of the curve.

Patients
Twenty-one patients were entered into this study. Their characteristics are shown in Table I. Their median age is 60 years (range 35-71) with a male-female ratio of 19:2. One patient suffered a relapse after lobectomy 1 year before. In the other patients no previous malignancies had been diagnosed. Eleven patients gained weight during treatment, seven had a stable weight and three lost weight. Performance score (ECOG) improved in ten patients by 1 point and 11 remained stable. There was one CR (confirmed by surgical resection), ten PRs, six SDs and four PDs.

Toxicity
Nineteen patients completed the treatment. In two patients at the last dose level of 25 mgm2 day-' carboplatin infusion was stopped on days 31 and 28. The first patient developed an infection at the VAP due to Staphyloccus aureus with neutropenic fever. After removal of the VAP leucopenia and temperature normalised within 5 days. In this patient radiotherapy was postponed for 1 week. In the other patient leucocytopenia grade IV and thrombocytopenia grade III led to stopping of carboplatin, but not of irradiation. Another complication of the VAP catheter was thrombosis of the subclavian vein 3 weeks after the end of the treatment (15 mg m2 day' x 31 days). Because of these complications, in the last three patients we used a percutaneously inserted venous catheter in the elbow with tip placement in the superior vena cava (Gesco Per-Q-Cath, San Antonio, Tx, USA). There were no complications.
Haemotological toxicity was dose limiting. Thrombocytopenia developed at the highest dose level in the fifth and sixth weeks of treatment, preceding leucocytopenia, which was prominent 1 week after the end of treatment. Both recovered within 2 weeks after completion of treatment. The nadir of leucocytes and platelets is given in Table II. No haematological toxicity was seen when carboplatin was given Non-haematological toxicity is described in Table III. Oesophagitis was mild and most prominent in the fourth week of treatment. It subsided in the second week after the end of treatment, except in two patients whose complaints disappeared in the third week. With a follow-up of 1 year no late oesophagitis had been encountered. The mean length of the oesophagus in the radiation field up to 40 Gy was 19 cm and thereafter 14 cm with no differences between oesophagitis WHO grades 0, 1 and 2. Acute radiation damage was also mild: dry cough and slight radiographic haziness in the radiation field from the fourth week on led to WHO grade II toxicity in all patients. Clinical features of pneumonitis such as fever did not develop. Corticosteroids were not used. Spinal cord damage and alopecia did not occur.

Lung function
Diffusion capacity corrected for alveolar volume (Kco) measured in 12 patients (the last three dose levels) before treatment was slightly decreased compared with age-and gender-matched controls: 77% (26) of predicted (s.d.). There was no change in Kco during and 2 weeks after treatment.

Discussion
Traditionally, thoracic radiotherapy is the main treatment for local inoperable NSCLC, despite the fact that its role is not factor; V,.p, pulmonary capillary blood volume. All measurements in individual patients were in duplicate and the mean is expressed as percentage of predicted ± s.d. (n = 12, from the last three dose levels). well defined (Kjaer, 1982. delivered in several ways tinuous treatment seems to hyperfractionation gives slil expense of more local toxic al., 1990). However, results distant metastases but also year follow-up are high (P rationale for adding a chem the local effect of radiation effects and may eradicate mi used before, but toxicity li (Bunn, 1992 oesophagitis has yet been encountered. A short-lived throm-,' * / --bocytopenia and leucopenia in the last week of treatment and a mild early oesophagitis were the most common problems. Attention should be paid to the venous access port to prevent /., infections and thrombosis of the subclavian vein. One option is to remove the VAP at the end of the treatment; another option is to use central venous catheters, which can be changed any time. Enhancement of radiation-induced lung dysfunction by carboplatin was not observed in the first 3 months. The ,' progressive fibrotic reaction probably counteracted shrinking of tumour tissue in the acute phase, keeping total lung function (TLC) steady during and shortly after treatment. Diffusion capacity remained unchanged. The initially measured pulmonary membrane factor and to a lesser extent 4 6 8 10 the capillary volume were consistently low in comparison rable platinum (g min-1-') with matched controls, which might be interpreted as a more extensive interstitial or small vessel involvement than could thrombocytopenia and AUC for be expected from CT scan of the chest alone. However, line is theg95o confidence interval during and after the combined treatment neither factors changed. Animal studies with irradiation showed inflammatory cells and protein leak in bronchoalveolar lavage fluid and in histological sections of the lung from 2 weeks after radiation exposure (Travis, 1980;Rosiello, 1993).

Bonomi, 1986). It has been
Interstitial and vascular changes could account for a persisto improve local control. Content low pulmonary membrane factor and decreased pulbe better than split course, and monary capillary volume (Collis and Steel, 1982;Collis, ghtly better local control at the 1982). Toxicity measured by ventilation rate and lethality in ,ity (Salazar et al., 1976;Cox et mice showed that addition of platinum-containing drugs to remain poor, mainly because of radiation gave no or hardly any acute damage (Von der because local relapse rates at 1 Maase et al., 1986;Tanabe et al., 1987;Steel, 1988). 'erez et al., 1987). This was the One of the problems with continuous carboplatin infusion otherapeutic agent that increases is that a progessively greater portion of UPt is accounted for by radiosensitising and cytostatic by catabolic products of proteins to which platinum has been icrometastases. Cisplatin has been covalently bound. The anisotropic, hydrophilic ultrafiltration imits its use in clinical practice membrane has a narrow pore size distribution. Membrane night be an alternative as it lacks pores have a diameter for molecules up to a molecular weight rotoxicity. Higher total carbop-of about 30000. Therefore, measurements of UPt represent with less toxicity by ambulatory not only free platinum but also platinum bound to different e may compensate for the slower small proteins. This shows the limitation of the described platin to its active form as com-method used for determination of platinum in the ultraor any schedule dependency of filtrate during continuous carboplatin infusion. However, the idiation (Micetich et al., 1985; fact that there is a relationship between the decrease in 1989). In vitro studies with proplatelets and AUC for ultrafilterable platinum (including proatin suggest an increased number tein products) demonstrates that the ultrafiltrate contains an ibits the same cytotoxic effects as active substance and is not just a non-specific measurement.
1985; Knox et al., 1986 the pharmacokinetics and thrombocytopenia associated with carboplatin found a linear relationship, subsequent analyses over wider ranges of carboplatin AUC have shown that this relationship is better described by a sigmoid Em.,, model defined by the Hill equation. This model was also applicable on the percentage decrease of platelets and AUC for ultrafilterable platinum during continuous infusion with carboplatin.
Local tumour control is determined by tumour size, radiation dose, tumour resistance factors, performance score and to a lesser extent the intrapulmonary localisation. Tumour size is important because sterilisation of tumour cells is only possible in small tumours (Dosoretz et al., 1993). Also chemotherapy achieves better results in smaller tumours. For radiobiological reasons split-course radiation incorporated into combination treatment is inferior to continuous radiation probably owing to repopulation or regeneration with resistant tumour cells during the split period. The effects of irradiation on local tumour control are difficult to assess because of an extensive fibrotic reaction around the shrinking tumour. Response rates of about 50% have been reported (Perez et al., 1982;Dosoretz et al., 1993). The response rate to carboplatin alone of NSCLC patients is low at around 10% (Kreisman et al., 1987;Bunn, 1992). In a phase I study with continuous carboplatin infusions alone 5/17 evaluable patients with advanced malignancies responded (Smit et al., 1991). However, pure local radiation treatment hardly influenced survival (Perez et al., 1982), while systemic treatment with platinum drugs offers slightly improved survival (Dillman et al., 1990;Schaake-Koning et al., 1992).
In conclusion, the optimal dose for continuous carboplatin infusion over 6 weeks is 20mgm-2day-' in combination with locoregional fractionated radiation therapy of 30 fractions of 2 Gy. Toxicity has been remarkably mild. A phase III study is necessary to evaluate survival benefit and clinical impact on local control and quality of life for a large group of inoperable stage III NSCLC patients.