Is high-dose intensity intraarterial cisplatin chemoradiotherapy for head and neck carcinoma feasible?

In this issue, Foote et al.1 report their preliminary experience in treating patients with advanced head and neck carcinoma using a combination of high-dose intraarterial cisplatin chemotherapy and concomitant, boost-accelerated radiation therapy. They conclude that this combination of treatment modalities is not feasible. This observation is based on 19 patients who received 3 or 4 weekly cycles of intraarterial cisplatin at doses of 150 mg/m² and simultaneous received intravenous sodium thiosulfate at a dose of 9 g/m² over 15–20 minutes followed by 12 g/m² over 6 hours. All patients received concurrent, concomitant, boost-accelerated radiation therapy consisting of 72 grays (Gy) in 42 fractions over 6 weeks. There were 2 patient deaths among the first 15 patients enrolled, both of which resulted from sepsis secondary to chemotherapy-induced neutropenia. Despite reducing the number of intraarterial infusions from four to three after the second death, the fourth patient who was treated subsequently on the modified protocol also developed neutropenia, sepsis, and thromboembolic events leading amputations. These observations led the authors to conclude that the combination of altered fractionation radiation therapy and intraarterial cisplatin in the doses outlined was not feasible.

In contrast to this conclusion, others, including me, believe that this treatment regimen is feasible. In the paragraphs below, I have outlined my personal experience in managing hundreds of patients treated with intraarterial cisplatin using the dose and route of administration similar to those used in this current study. I also have highlighted the published experience of other investigators, all of whom concluded that the high-dose-intensity approach is feasible and highly efficacious.

First, it must be asked: Why intraarterial chemotherapy when intravenous regimens are available? Oncologists are aware of the pharmacokinetic advantages of administering chemotherapy through an intraarterial route. The first pass of the drug is through the tumor bed, allowing exposure of a high concentration. Patients with head and neck tumors are quite suitable for intraarterial chemotherapy because of easy access to the arterial supply and the relative advantage of the slower blood flow to the head and neck region. Although intraarterial chemotherapy for head and neck carcinoma has been investigated for almost 5 decades, it was the development of interventional endovascular techniques that paved the way for the accurate delivery of drugs to targeted sites with minimal complications. Capitalizing on these technologic advances, we began treating patients with advanced head and neck carcinoma in 1988 at the University of California, San Diego. Our protocol was unique, in that sodium thiosulfate was combined with the intraarterial infusion, which allowed the administration of an extremely high dose intensity of cisplatin to patients with minimal major toxicity.2 In a Phase I study, we determined that cisplatin could be administered at a dose intensity of 150 mg/m² per week up to a total dose of 600 mg/m² over 22 days. Higher doses resulted in a reversible nephrotoxicity in which patients lost large quantities of electrolytes. Our subsequent Phase II study of radiation plus cisplatin, which we nicknamed RADPLAT, incorporated the unique high-dose cisplatin drug-delivery technique with radiation therapy.3 In that study, which was carried out at the University of Tennessee, Memphis, we demonstrated that the combination of radiation therapy (1.8–2.0 Gy per day in 35 fractions over 7 weeks) combined with 3 or 4 weekly infusions of cisplatin was highly efficacious and safe. Among the 57 patients in that study who were evaluable for response, 91% of patients had a complete response, and the estimated survival rate at 1 year was 67% (median follow-up, 18 months; range,12–26 months). Overall, there were 32 events of Grade 3–4 toxicity (according to the National Cancer Institute [NCI] scale) related to treatment that affected 25 of 60 patients. There were no treatment-related deaths. Encouraged by these results, we continued to treat all eligible patients with RADPLAT over an 8-year interval while prospectively maintaining a data bank on all patients. An analysis of this cumulative experience was reported in which 213 patients comprised the study group.4 We observed a complete response rate in the primary site of 80% for all patients. The projected rate of locoregional disease control was 78%, and the projected rate of overall survival at 5 years was 38%. The toxicity profile was as follows: Seventy-two patients experienced 1 episode of Grade 3–4 toxicity; 8 patients experienced 2 episodes of Grade 3–4 toxicity; and 2 patients experienced 3 episodes of Grade 3–4 toxicity;. This included seven central nervous system events, including five cerebrovascular accidents and two transient ischemic attacks. All patients subsequently were able to remain ambulatory and physically active, even though there were residual motor deficits. Although there were 6 episodes of Grade 5 toxicity (2.8%), this rate was not unusually high compared with other studies. These results in a highly unfavorable group of patients indicated the potential for improved outcome as measured by organ preservation, local regional disease control without salvage surgery, and (possibly) overall survival.

RADPLAT subsequently was put to the test of a multicenter trial.5 This was an NCI-sponsored study that was done through the Radiation Therapy Oncology Group (RTOG) in which 67 patients with T4 disease were enrolled at 10 centers in the United States. An important question of the trial (RTOG Trial 9615) was whether this highly technical approach could be used safely in a multicenter setting while reproducing the efficacy results. Seven of the 10 participating centers did not have any previous experience with concomitant intraarterial chemoradiation, which allowed for a comparison with the experienced sites. The data showed that the rate of toxicity for the inexperienced centers was higher. However, the overall delivery rate for treatment did not differ significantly between the two groups. Although two toxic deaths were encountered, this did not warrant stopping the study prematurely. Relevant to the high number of catheter-related complications in the current study, very few were encountered. Surprisingly, the inexperienced centers had a success rate with catheterizations similar to the rate achieved at the experienced sites. It is possible that the workshop held prior to opening the study may have helped to circumvent this pitfall.

Intraarterial chemotherapy currently is being studied by many centers in several countries. At the Netherlands Cancer Institute, the results from a completed Phase II study using the RADPLAT protocol recently were published.6 In that study, there were 61 patients with T4 lesions and 18 patients with T3 lesions, whereas 49 patients had lymph node-positive (N+) disease. A complete local tumor response was achieved in 72 patients (91%), and a partial response was achieved in 3 patients. The complete response rate of cervical lymph node metastases was 90%. The 1-year and 2-year locoregional control rates were 82% and 69%, respectively. The median overall survival was 2.2 years, with a 3-year overall survival probability of 43%. Acute toxicities were as follows: Grade 3 and 4 hematologic toxicity, 22% and 16%, respectively; Grade 3 and 4 nephrotoxicity, 0%; Grade 3 mucositis, 43%; Grade 3 skin reactions, 24%; Grade 3 toxicity of the upper gastrointestinal tract, 57%; Grade 3 nausea, 20%; and Grade 3 subjective hearing loss, 10%. There were no Grade 4 nonhematologic toxicities. The Grade 5 toxicity rate (treatment-related deaths) was 3.8%. Six of 33 patients (18%) who achieved complete remission required tube feeding 2 years after treatment without intercurrent salvage surgery. The investigators in that study concluded that supradose, superselective, intraarterial cisplatin and concomitant radiation is an effective organ-preserving therapy in an unfavorable group of patients. They also concluded that the regimen is justified in unresectable patients despite the substantial toxicity. The Dutch are now conducting a randomized trial comparing RADPLAT with intravenous cisplatin chemoradiation in a study that is being conducted at four centers throughout the Netherlands. Other intraarterial protocols for advanced head and neck carcinoma have been carried out in the United States,7, 8 Japan,9–12 Taiwan,13 Germany,14 and Italy.15 In Japan, where intraarterial chemotherapy for paranasal sinus carcinoma has been used for many years,11, 12 investigations recently have expanded to include other upper aerodigestive tract sites.13, 14 Yokoyama reported the use of neoadjuvant intraarterial cisplatin with sodium thiosulfate neutralization for 97 patients with advanced lesions, resulting in a complete response rate of 72% and preservation of the larynx in all patients.13 Furatani et al. treated 39 patients who had tongue carcinoma with intraarterial carboplatin and concurrent radiation therapy followed by 2 courses of systemic chemotherapy with cisplatin and 5-fluorouracil (5-FU). In their study, 33 patients had oral tongue carcinoma, and 6 patients had tongue base carcinoma. Although 9 patients in that study had recurrent disease, their overall survival rate at 3 years was 58.9%.12 Kovacs et al. treated 73 patients who had operable oral and oropharyngeal carcinoma with neoadjuvant intraarterial cisplatin/systemic sodium thiosulfate neutralization and radical surgery followed by concurrent radiation and chemotherapy weekly with docetaxel.14 The cumulative survival at 4 years was 74%. A common theme of all centers was that the results strongly supported continued investigation of the intraarterial drug-delivery approach.

I only can speculate why the Mayo Clinic group encountered unacceptable toxicity in their pilot trial. It is perplexing that as many as three patients could develop catastrophic events from severe neutropenia. Neutropenia, which is common to all chemoradiation protocols for head and neck carcinoma, is to be expected. The proactive use of growth factors can minimize the risk of Grade 4 neutropenia and avoid sepsis. The maintenance of nutritional status is important during therapy. The Mayo Clinic group noted that only 3 of 17 patients had percutaneous endoscopic gastrostomy (PEG) tube placement prior to the start of therapy, whereas 12 patients required PEG tube placement after the start of therapy. In our center, most patients receive PEG tubes prior to the initiation of therapy for the reason that poor nutritional status may have an impact on the outcome and the susceptibility for toxicity.

The number of catheter-related problems encountered also is uncharacteristic and definitely is not the experience of others. At our own center, interventional radiology routinely performs intraarterial screening studies to confirm the feasibility of “RADPLAT” therapy in each patient. Similar to the RTOG 9615 trial, it would seem prudent that institutions undertaking intraarterial protocols for the first time undergo in-service training. This was done for the NCI-sponsored multicenter trial (RTOG 9615), the Dutch study, and most other studies. Although the interventional techniques used are standard to the practice of interventional radiology, patients with advanced head and neck carcinoma represent a more fragile group.

Regardless of toxicity and feasibility, the debate will continue over the use of intraarterial chemotherapy and whether it has any practical advantage over intravenous regimens. Kumar et al. compared the outcome results from RTOG 9615 with the results from RTOG 9703, which was a Phase II randomized trial that compared 3 concurrent chemoradiation protocols for advanced head and neck carcinoma.16 All of the regimens in the latter trial used standard-fractionation radiation therapy (70 Gy over 7 weeks in Arms 1 and 3; 70 Gy every other week over 13 weeks in Arm 2) with daily intravenous cisplatin (10 mg/m²) and 5-FU (400 mg/m² per day as a continuous infusion during the last 2 weeks of radiotherapy) for Arm 1; and, every other week, with intravenous 5-FU (800 mg/m² per day for 5 days) and oral hydroxyurea (1 g twice daily for 6 days) for Arm 2; and weekly intravenous cisplatin (20 mg/m²)/taxol (30 mg/m²) for Arms 3. The estimated 2-year local control and overall survival rates for patients with T4 disease were as follows: RTOG 9703 Arm 1, 43.1% and 47.9%, respectively; RTOG 9703 Arm 2, 50% and 52%, respectively; RTOG 9703 Arm 3, 53.5% and 56%, respectively; and RTOG 9615, 53.9% and 58.8%, respectively. Using a Cox proportional hazards model stratified by recursive partitioning analysis class, RADPLAT therapy reduced death rates by 21% compared with Arm 1 on RTOG Protocol 9703; however there was virtually no reduction in death rates compared with Arms 2 and 3 of RTOG Protocol 9703 (0.8% and 1.8%, respectively). Thus, RADPLAT therapy appears comparable to contemporary intravenous chemoradiation regimens and warrants further testing in a Phase III randomized trial comparing RADPLAT therapy with intravenous chemotherapy. Even though toxicity is common to all of these protocols, the highly active antitumor effects of the intraarterial approach and the demonstration of its feasibility for use in almost every center warrant its continued investigation. Although the high toxicity encountered by the Mayo Clinic group is disappointing, it is inconsistent with other reports.