Prior chemoradiotherapy adversely impacts outcomes of recurrent and second primary head and neck cancer treated with concurrent chemotherapy and reirradiation

Abstract

BACKGROUND:

It has been shown that concomitant chemotherapy (C) with reirradiation (ReRT) is feasible and effective for select patients with recurrent or second primary head and neck cancer (HNC). To examine potential prognostic factors associated with survival, the authors of this report retrospectively reviewed the outcomes of patients who received CReRT.

METHODS:

The study cohort comprised previously irradiated patients with nonmetastatic disease from 9 consecutive phase 1 and 2 protocols for poor-prognosis HNC. For all patients, reirradiation (ReRT) was delivered with concurrent chemotherapy. Chemotherapy generally was 5-fluorouracil, hydroxyurea, and a third agent.

RESULTS:

One hundred sixty-six patients were identified, including 81 patients who underwent surgical resection or debulking before enrollment. The median ReRT dose was 66 gray. After a median follow-up of 53 months among surviving patients, the median overall survival (OS) was 10.3 months. The 2-year rates for OS, disease-free survival, locoregional control, and freedom from distant metastasis were 24.8%, 19.9%, 50.7%, and 61.4%, respectively. Thirty-three patients (19.9%) died of treatment-related toxicity. In subgroup analysis, survival was significantly reduced in patients who received previous concurrent chemoradiotherapy (CRT) compared with patients who were naive to CRT (2-year OS rate, 10.8% vs 28.4%; P = .0043). In multivariable analysis, prior CRT was associated independently with OS along with surgery before protocol treatment, full-dose ReRT, and radiotherapy interval.

CONCLUSIONS:

CReRT achieved a long-term cure for a small group of patients with recurrent or second primary HNC. Previous treatment with CRT was among the important prognostic factors for survival. Because of the associated risk of severe toxicity, CReRT should be limited only to carefully selected patients. Cancer 2011;. © 2011 American Cancer Society.

Despite significant improvements in the treatment of locoregionally advanced head and neck cancer (HNC), locoregional recurrence continues to be problematic, occurring at a rate of 20% to 30%.1-4 When recurrences develop in a previously irradiated area, treatment options are limited, and the prognosis is poor. In addition, patients with HNC are at risk for second primary malignancies, many of which are in the head and neck region, at a rate of up to 3% per year.5-8 For both of these scenarios, surgical salvage is the preferred treatment approach, but only a few patients are candidates for radical surgery because of tumor extent and medical comorbidities.9, 10 Patients typically are managed with systemic chemotherapy, but this approach is associated with a median survival of only about 6 to 8 months, and a cure is unlikely.11-13

A second course of radiotherapy (RT) or reirradiation (ReRT) is a potentially curative therapeutic option. Although the risk of severe toxicity is significant, it has been demonstrated that ReRT is feasible, and long-term survival can be achieved in 25% to 45% of select patients.14-16 Tumors that recur or persist despite a course of RT suggest that there are radioresistant clonogens.17 To overcome such resistance, protocols that combine radiosensitizing chemotherapy with ReRT have developed, and this approach has demonstrated better long-term survival compared with ReRT alone.18

At the University of Chicago, concurrent chemotherapy with ReRT (CReRT) for recurrent or second primary HNC has been investigated over the last 2 decades in a series of phase 1 and 2 protocols. Most protocols were based on the FHX platform, consisting of 5-fluorouracil (5-FU), hydroxyurea (HU), and radiotherapy (RT). The rationale for this combination was based on independent cytotoxic activities and radiosensitizing properties of 5-FU and HU.19 Studies from our group20-22 as well as several other groups18, 23-25 have demonstrated long-term survival with FHX-based regimens, and the reported 2-year overall survival (OS) rates range from 15% to 45%.

In the primary treatment of locoregionally advanced HNC, concurrent chemoradiotherapy (CRT), as opposed to RT alone, improved locoregional control (LRC) and survival in both the definitive and postoperative settings, and the use of CRT has become more common in recent years.1, 2, 26, 27 Consequently, the nature of recurrent tumors may have changed. Tumors that recur despite more aggressive CRT may be more therapy-resistant than those described in the earlier studies, most of which were recurrences after patients received RT alone. For tumors that recur after CRT, retreatment with CReRT may be less effective. For the current report, we have updated our experience with treating recurrent and second primary HNC with CReRT, and we have specifically investigated the efficacy of CReRT in patients whose initial course of therapy was CRT.

MATERIALS AND METHODS

Patients

Between 1986 and 2006, 382 patients who had HNC with a poor prognosis were treated on 9 consecutive phase 1 and 2 protocols that investigated CRT.21, 28-35 From this cohort, we included nonmetastatic patients who had recurrent or second primary squamous cell carcinoma in a previously irradiated region for this analysis.

Treatment Protocols

The details of each protocol were reported previously21, 28-35 and are summarized in Table 1. Most protocols used 14-day CReRT cycles, which consisted of 5 consecutive days of CReRT followed by 9 days of no treatment. Chemotherapy generally was 5-FU and HU with a third agent (Table 1). A complete course was comprised 4 to 7 cycles based on the extent of preprotocol surgery as well as ReRT fractionation. One protocol (13881) included pemetrexed-based induction chemotherapy before CReRT, and, in that protocol, CReRT was administered continuously without breaks.33

Table 1. Protocol Summary

Reference	Protocol	Years Open	No. of Patients	Chemotherapy	Total RT Dose (Fraction Size). Gy
RT indicates radiotherapy; Gy, gray; HU, hydroxyurea; 5-FU, 5-fluorouracil; BID, twice daily; Gem, gemcitabine; Pem, pemetrexed.
Haraf 199129	86N2	1986-1988	14	HU, 5-FU	50 (1.8-2 daily)
Haraf 199230	5128	1987-1989	3	HU, 5-FU, cisplatin	70 (2 daily)
Vokes 199434	5522	1989-1991	10	HU, 5-FU, cisplatin	50-70 (2 daily)
Vokes 199434	6094	1991-1993	10	HU, 5-FU, cisplatin	50-75 (1.5 BID or 2 daily)
Brockstein 200028	7034	1993-1997	42	HU, 5-FU, paclitaxel	60-75 (1.5 BID)
Salama 200531	9071	1998-2001	10	HU, 5-FU, irinotecan	60-75 (1.5 BID)
Milano 200521	9105	1998-2001	26	5-FU, paclitaxel, Gem	60-75 (1.5 BID)
Seiwert 200832	33B	2001-2004	28	HU, 5-FU, bevacizumab	60-72 (1.8-2 daily)
Villaflor 200833	13881	2005-2006	23	Pem, Gem (induction); Pem, carboplatin (with RT)	60-70 (1.8-2 daily)

All patients were evaluated initially by a multidisciplinary team consisting of medical oncologists, head and neck surgeons, and radiation oncologists. Patients for whom surgical resection was not feasible proceeded directly to CReRT (except for patients in Protocol 13881, for whom induction chemotherapy was initiated). After an analysis suggested that surgical debulking might improve outcomes, surgical debulking or cytoreduction was encouraged for all patients before CReRT.

Reirradiation Guidelines

Radiation planning and delivery techniques have been discussed in detail previously.22 All patients underwent an RT simulation procedure with appropriate immobilization. Whenever available, treatment plans from the initial course of RT were evaluated to define previously irradiated areas and maximal doses to critical structures. The minimally acceptable treatment volume was the gross tumor, as identified on physical examination and imaging studies, with at least a 1-cm margin. Smaller margins were used when the cumulative dose to the spinal cord or brainstem became prohibitive. Elective treatment of adjacent sites or draining lymph nodes generally was avoided.

The recommended RT doses varied initially, but most patients were treated according to the following guidelines: 60 gray (Gy) to microscopic disease, 66 to 74 Gy to gross disease, and 72 to 76 Gy to bulkier disease (>4 cm). The lifetime spinal cord dose was limited to 50 Gy. Most patients received 2.0 Gy daily or 1.5 Gy twice daily with a 6-hour interfraction interval. Two patients received an Iridium-192 brachytherapy boost, and 2 others received intraoperative RT.

Follow-Up

Patients were followed by the multidisciplinary team with comprehensive physical examinations and computed tomography scans 1 month after treatment and every 3 to 6 months thereafter. Complete metabolic panels and complete blood counts were obtained at every medical oncology visit. Thyroid function tests, swallowing function studies, and endoscopy were obtained every 6 months or as clinically indicated. Clinical outcomes, including treatment response, complications, disease recurrence, and death, were recorded prospectively.

Statistical Analysis

The durations of OS, disease-free survival (DFS), LRC, and freedom from distant metastasis were calculated from the first day of protocol therapy. Long-term complications were graded using the Common Terminology Criteria for Adverse Events, version 3.0. Actuarial survival curves were generated using the Kaplan-Meier method, and comparisons were made with the log-rank test. Variables that had P values ≤ .10 were tested further in multivariable analysis using a Cox proportional hazards model. All data were analyzed using the JMP 9 for Windows software program (SAS Institute, Inc., Cary, NC).

RESULTS

Patient, Disease, and Treatment Characteristics

One hundred sixty-six patients met our study inclusion criteria. The median age was 59 years (range, 30-83 years). Most patients had recurrent disease, but 19 patients (11.4%) presented with second or third primary tumors. Other patient and disease characteristics are listed in Table 2.

Table 2. Patient Characteristics

Characteristic	No. of Patients (%)
ECOG PS indicates Eastern Cooperative Oncology Group performance status; PY, pack-years.
All patients	166 (100)
Median age [range], y	59 [30-83]
Sex
Men	117 (70.5)
Women	49 (29.5)
Second/third primary	19 (11.4)
ECOG PS
0	39 (23.6)
1	93 (56)
2	31 (18.8)
3	3 (1.8)
Smoking history, PY
None	34 (20.5)
Pipe/cigar/chew	10 (6)
<20	9 (5.4)
20-40	40 (24.1)
>40	72 (43.4)
Unknown	1 (0.6)
Alcohol consumption, no. of drinks
None	44 (26.5)
<1/wk	20 (12)
≤1/d	27 (16.3)
>1/d	62 (37.3)
Unknown	13 (7.8)
Primary disease site
Oral cavity	60 (36.1)
Oropharynx	45 (27.1)
Nasopharynx	4 (2.4)
Hypopharynx	10 (6)
Larynx	30 (18.1)
Supraglottic larynx	2 (1.2)
Sinus	4 (2.4)
Thyroid	1 (0.6)
Other	10 (6)

Eighty-one patients (48.8%) underwent surgical resection or debulking before CReRT (Table 3). The median ReRT dose was 66.0 Gy. Seventy-three patients (44%) had previous systemic chemotherapy exposure, including 40 patients (24.1%) who received CRT.

Table 3. Treatment Summary

Treatment	No. of Patients (%)
RT indicates radiotherapy; ReRT, reirradiation; Gy, gray.
Surgical resection	81 (48.8)
Type of surgery
Radical	58 (71.6)
Debulking	19 (23.5)
Unknown	4 (4.9)
Median Previous RT dose, Gy	68.4
Median ReRT dose [range], Gy	66.0 [6.0-75.2]
Median Lifetime RT dose [range], Gy	131.0 [57.2-207.6]
Median RT interval [range], mo	18.0 [1.2-482.5]
RT fractionation
Once daily	93 (56)
Twice daily	73 (44)
Previous chemotherapy
Any exposure	73 (44)
Concurrent with RT	40 (24.1)

Toxicity

The toxicities associated with therapy were substantial, as expected. All grade 4 and 5 toxicities are presented in Table 4. Thirty-three patients (19.9%) died from treatment-related complications, including 18 patients who died during therapy and 15 patients who died after completing protocol therapy. It is noteworthy that patients who had received previous CRT tolerated CReRT very poorly, and 12 of 40 such patients experienced grade 5 toxicity. There were 15 episodes of carotid hemorrhage, 10 of which were fatal. Eighteen patients developed osteoradionecrosis and required surgical intervention. There were no episodes of osteoradionecrosis when the lifetime RT dose was <120 Gy. Despite a median lifetime dose of 131 Gy in the cohort, neurologic complications were rare. There was 1 episode of spinal cord myelitis and another of brachial neuropathy. There was no episodes of brainstem necrosis.

Table 4. Toxicity

Complication	No. of Patients (%)
Treatment-related death	33 (19.9)
Carotid hemorrhage	15 (9)
Osteoradionecrosis	18 (10.8)
Myelopathy	1 (0.6)
Peripheral neuropathy	1 (0.6)

Swallowing function assessment was available for 102 patients (61%). Among these, 26 patients (25%) were able to swallow a solid diet without difficulty; 7% used a gastrostomy tube to supplement <50% of their diet, and 56% depended on gastrostomy for their entire diet. At last follow-up, 68% of patients had a gastrostomy tube in place.

Of 84 patients who were evaluated for speech function, 21 patients (25%) had normal speech according to their physicians, and 49 patients (58%) more had intelligible speech. Six patients lost the ability to speak after CReRT. The remaining 8 patients were surgically mute.

Treatment Response and Disease Outcomes

At the completion of protocol therapy, 134 patients were assessed for initial treatment response. Of these, 31 patients (23%) underwent surgical removal of all measurable disease before CReRT, and their treatment response could not be determined. Of the remaining 103 patients, 72 achieved a complete response (70%), and 15 had a partial response (15%). Five patients had stable disease, and 11 patients developed progressive disease during CReRT.

At a median follow-up of 53.0 months (range, 7.0-188.4 months) for the patients who remained alive, the median survival was 10.3 months (95% confidence interval [CI], 8.4-12.6 months). The 2-year and 5-year OS rates were 24.8% (95% CI, 18.5%-31.7%) and 14.3% (95% CI, 9.1%-20.7%), respectively (Fig. 1). The 2-year and 5-year DFS rates were 19.9% (95% CI, 14.2%-26.6%) and 13.4% (95% CI, 8.4%-19.8%), respectively.

Figure 1. Overall survival is illustrated for all patients (n = 166).

Ninety patients progressed (54.2%). Of those who progressed, 40 patients (24%) had only locoregional progression, 27 patients (16%) had only distant progression. The remaining patients had both locoregional failures and distant metastasis. The 2-year LRC and freedom from distant metastasis rates were 50.7% (95% CI, 41.3%-60.5%) and 61.4% (95% CI, 51.5%-70%), respectively.

Analysis of Prognostic Variables and Subgroups

To identify potential prognostic factors associated with survival, univariate and multivariable analyses were performed. In univariate analysis, previous CRT, surgery before protocol, ReRT dose, RT fractionation, and administration of a third chemotherapeutic agent were associated significantly with survival; whereas performance status, age, recurrence versus second primary, treatment decade (1986-1996 vs 1996-2006), and RT interval were not. When those factors that had P values ≤ .10 were tested further in multivariable analysis, previous CRT, surgery before protocol, and ReRT dose retained a significant association with survival (Table 5). In addition, the RT interval became significant.

Table 5. Multivariable Analysis of Overall Survival

Variable	HR [95% CI]	P
HR indicates hazard ratio; CI, confidence interval; Ref, referent category; CRT, concurrent chemoradiotherapy; RT, radiotherapy; Gy, gray.
Surgery before protocol
No (Ref)	1.00	.0006
Yes	0.52 [0.36-0.76]
Previous CRT
No (Ref)	1.00	.0043
Yes	1.83 [1.21-2.70]
RT dose, Gy
<60 (Ref)	1.00	<.0001
≥60	0.35 [0.23-0.53]
RT interval, mo
<36 (Ref)	1.00	.0259
≥36	0.64 [0.42-0.95]
RT fractionation
Daily (Ref)	1.00	.0662
Twice daily	0.71 [0.49-1.02]
Performance status
>1 (Ref)	1.00	.8760
0/1	1.04 [0.66-1.68]
Third chemotherapy agent
No (Ref)	1.00	.9751
Yes	0.99 [0.52-1.78]

By using the 4 prognostics factors that were associated significantly with survival, as indicated above, an exploratory subgroup analysis was performed. In 11 patients who had all 4 favorable factors (ie, no previous CRT, surgery before protocol, ReRT dose ≥60 Gy, and RT interval ≥36 months), the 2-year OS rate was 63.6%, and 7 of 11 patients remained alive at the last follow-up. Conversely, for the 11 patients who had received previous CRT and <60 Gy of ReRT, the median survival was only 1.2 month, and there were no 2-year survivors. When patients were stratified into 3 risk groups according to the number of risk factors (0-1, 2, or 3-4 risk factors), OS differed significantly between the risk groups (P < .0001) (Fig. 2).

Figure 2. This graph illustrates overall survival stratified by the number of risk factors.

Survival Outcomes for Patients Previously Treated with CRT

Forty patients (24%) whose initial therapy consisted of RT with any concurrent chemotherapy were analyzed separately. Among them, 11 patients initially received FHX-based regimens, and 29 patients received other chemoradiotherapy platforms. In this group, the median survival was 7.3 months, and the 2-year OS rate was 10.8%. These outcomes differed statistically compared with the outcomes of patients who had not received CRT (median survival, 12.2 months; 2-year OS rate, 28.4%; P = .0043). The median DFS was 4.0 months, the 2-year DFS rate was 12%, and these outcomes also were worse than those for patients who were naive to CRT (P = .0008). Only 5 of 40 patients lived beyond 2 years. Survival outcomes were not significantly improved in those who underwent surgery before CReRT (n = 22; 2-year OS rate, 13.6%). Any previous exposure to chemotherapy (n = 73) also was associated with a worse median survival (8.1 months vs 12.2 months) and 2-year OS rate (21.8% vs 27.1%), with a difference that trended toward statistical significance (P = .0895).

DISCUSSION

For patients with recurrent or second primary HNC in a previously irradiated area, we and others have demonstrated that CReRT is feasible and provides the chance of a cure in a subset of patients.18, 22, 25, 36 In the current study, we present the long-term outcomes of 166 patients who received CReRT. Similar to other reported studies, we demonstrated that 2-year survival could be achieved in 25% of patients.18, 22, 36 However, with this treatment approach, the risk of severe toxicity was substantial, including 20% treatment-related deaths. The rate of severe toxicity approaching that of 2-year survival highlights the need to improve therapeutic efficacy and patient selection criteria.

Several studies have proposed pretreatment prognostic factors, such as tumor volume, resectability, RT interval, and the presence of comorbidities, which may help select optimal patients for CReRT.22, 37-43 In this study, we propose another potential prognostic factor, which is previous treatment with CRT. It is believed that the addition of chemotherapy to radiation provides synergy or sensitization and intensifies the effect of RT.44 Therefore, if a tumor recurs despite such intensified therapy, then it likely has acquired therapeutic resistance, and subsequent treatments may not be as effective. Consistent with this hypothesis, we demonstrated that survival after CReRT was reduced significantly if the patient initially received CRT. With CRT used increasingly for definitive and postoperative management of patients with HNC, now, we may be facing recurrent tumors that are more therapy-resistant than in the past. It is interesting to note that, despite the great advances in chemotherapy and RT, the survival outcomes after CReRT in our recent protocols are not much better than those from the initial years of our experience. Novel strategies, such as integration of molecularly targeted agents, hypoxic cell sensitizers, and viral oncolytic therapy, may be necessary to mitigate the observed therapeutic resistance.32, 45-47

Although prior treatment with CRT was associated significantly with poor survival in the multivariable analysis, it still is possible that there may be confounding variables. For example, the stage of disease at initial presentation was not analyzed. It is possible that patients with more extensive initial disease were managed with CRT instead of RT alone, and perhaps the recurrent disease also was more extensive in these patients, resulting in a poor outcome. The current study did not have enough power to determine whether the initial CRT platform influenced response to the CReRT administered. The majority of patients who had received CRT were treated on non-FHX, platinum-containing regimens. Therefore, we do not know whether patients who receive initial treatment with a platinum agent or treatment with the FHX platform are particularly resistant to subsequent therapy. In addition, it is important to note that patients who received previous CRT tolerated CReRT very poorly, with 30% experiencing grade 5 toxicity. The high rate of treatment-related deaths also contributed to the poor OS observed.

In this study, we observed that surgical resection before CReRT was associated with improved survival. It is conceivable that removing most radioresistant clones by surgery increases the likelihood of tumor eradication with CReRT. However, this result must be interpreted with caution considering the inherent biases in patient selection. Furthermore, operating in previously irradiated and surgically manipulated tissues requires some level of expertise. When we initiated this study, surgical cases were mainly low-volume tumors that could be resected completely. More recently, we have been performing tumor debulking when possible as a form of cytoreduction. There is evidence that, after surgery, growth factors are released that may promote tumor recurrence.48 The benefit of cytoreduction must be weighed against potential for tumor regrowth after surgery, especially when surgery is not a radical, en-bloc resection. It also is noteworthy that there was no significant survival difference in this series whether the surgery was radical or debulking (data not shown).

The major concern with CReRT is the risk of irreparable damage that can impact quality of life or even survival. With regard to speech and swallowing functions, our data suggest that functional speech usually is maintained after CReRT. Swallowing function, however, was more problematic, and >66% of our patients required gastrostomy for feeding. The use of advanced RT delivery techniques, such as intensity-modulated RT or image-guided RT, have demonstrated feasibility and efficacy in the setting of head and neck ReRT.39, 49, 50 The use of these techniques may lead to further improvement in quality-of-life outcomes.

The results from this study are subject to limitations from the retrospective nature of the analysis, and they need to be considered as only hypothesis-generating. The most important limitation was the inability to evaluate and account for preprotocol treatments, including the initial disease characteristics and preprotocol treatments received. The extent of surgery or adverse pathologic features was not controlled in the current analysis. Conversely, to our knowledge, this study represents the largest series of head and neck CReRT reported to date. Patients were treated relatively consistently under the guidance of protocols. Furthermore, data were collected in a prospective manner through clinical trials with meticulous follow-up.

In conclusion, for patients with recurrent HNC in a previously irradiated area, CReRT is a treatment option that offers the small but real chance of a cure in select patients. Previous treatment with CRT was an important pretreatment characteristic and was associated with poor survival after CReRT. Because of the risk of severe toxicity, the use of CReRT should be limited to investigational protocols or experienced centers.

FUNDING SOURCES

This study was supported in part by the Valda and Robert Svendsen Foundation, and the sponsor had no involvement in the study design; in the collection, analysis and interpretation of data; in the writing of the article; or in the decision to submit the article for publication.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.