A comparative analysis between sequential boost and integrated boost intensity-modulated radiation therapy with concurrent chemotherapy for locally-advanced head and neck cancer

Radiation therapy for locally advanced squamous cell carcinoma of the head and neck (LAHNC) has changed considerably over the past two decades with the advent of intensity modulated radiation therapy (IMRT). While the toxicity profile for patients with LAHNC has improved, significant acute and late toxicities remain. In the pre-IMRT era, definitive treatment consisted of conventional radiation at a constant dose-per-fraction with successive narrowing of fields to account for normal tissue tolerances. With newer IMRT techniques, one can differentially dose gross disease, high-risk subclinical, and low-risk subclinical disease without changing the overall treatment volume. This planning technique gained popularity because of improved planning efficiency (only a single inverse treatment plan is required), and early studies suggesting improved dose distributions [1, 2]. As a result, this method of IMRT, often referred to as simultaneous integrated boost (SIB), became the predominant mode of delivering radiation for LAHNC.

Studies have shown that dose-per-fraction to gross disease as high as 2.2 Gy/fraction with concurrent chemotherapy [3] and 2.4 Gy/fraction with radiation alone [4] with integrated boost are safe and effective. Conversely, a dose-per-fraction as low as 1.6 Gy/fraction shows equivalent tumor control to the more conventional 2.0 Gy/fraction [5]. Despite this working range, there has been variability among integrated boost regimens with regard to toxicity and outcome profiles [6].

As inverse treatment planning techniques continue to improve, a renewed interest in sequential boost planning has developed. To date, few studies directly examine the differences between sequential boost and SIB IMRT in the treatment of LAHNC. Purely dosimetric studies on limited patient numbers or patient surrogates have varied regarding the advantage of one technique over the other. Older studies suggested that SIB improved normal tissue sparing and/or improved conformality [1, 7, 8], while recent studies demonstrate reduced dose to normal structures in the neck, and improvement in conformality with sequential boost IMRT [9, 10]. Studies comparing clinical outcomes and toxicities among patients treated with either technique are lacking. A recent randomized study from Thailand explored toxicities and outcomes in a small cohort of nasopharyngeal carcinoma patients and showed no difference in acute toxicity rates or early outcomes between the two IMRT techniques [11]. However, the absence of more common head and neck cancer subtypes in this study and the low overall rate of grade 3 or higher toxicity compared to historical controls may limit the applicability of these results.

At our institution, a transition from SIB to sequential boost IMRT for LAHNC was intentionally made for all patients. This was based on clinical observations that many SIB patients developed acute radiation toxicities requiring treatment breaks or premature treatment completion and sequential boost was attempted to mitigate these effects. In this study, we evaluate the toxicities, survival, and recurrence rates associated with sequential boost IMRT versus SIB in consecutive cohorts of patients using similar concurrent chemotherapy.

The widespread use of IMRT for definitive treatment of LAHNC has resulted in an overall move from sequential boost to integrated boost radiation planning. This shift was influenced by early dosimetric studies suggesting improved dose distribution, initial ease of designing one plan versus two or more, and the predominance of using the SIB technique in Cooperative Group studies [14, 15]. At our institution, we made a conscious shift away from SIB to sequential boost IMRT for LAHNC based on our clinical observation that patients treated with SIB experienced more severe acute skin and pharyngeal toxicity and seemed less likely to complete treatment. Here, we systematically examined the outcomes and acute toxicities of comparable cohorts of consecutive patients who received definitive chemoIMRT for LAHNC either with SIB or sequential boost techniques. This is the first study comparing IMRT techniques among relatively large cohorts that include the more predominant head and neck subsites and were consecutively treated using similar chemotherapy and dose-volume constraints. Overall, we find that these techniques appear equivalent with respect to treatment outcomes though sequential boost IMRT is associated with some improvement in rates of acute toxicity.

Specifically, dysphagia was consistently improved in the sequential boost cohort. Dermatitis and mucositis rates were also improved, however statistical significance was variable depending on the group or sub-group analyzed. For instance, among the entire sequential boost cohort, rates of high-grade dermatitis were improved and high-grade mucositis were statistically comparable to SIB, while among only patients that completed most of their concurrent chemotherapy, this was reversed. The paradoxical effect with mucositis slightly improving (70% vs 68% grade 3/4) and dermatitis worsening (58% vs 66% grade 3/4) as more cycles of concurrent chemotherapy were completed in the sequential boost cohort is, in part, the consequence of chemotherapy being held for high-grade mucositis. Patients receiving less chemotherapy for this reason will of course have worse mucositis which prompted the initial decision to hold systemic therapy, but may also have improved dermatitis as a consequence. Additionally, the severity of dermatitis (more so than dysphagia and mucositis) can vary depending on factors independent of treatment volumes and disease, such as body habitus.

These differences in acute toxicity do not appear to be the results of variable dose and dose-volume parameters to key structures between SIB and sequential boost treatments. Instead, we postulate that the observed improvement in dermatitis and dysphagia seen clinically with sequential boost is due to decreasing radiation exposure to the lower neck and pharyngeal constrictors from 7 to 5 weeks. As studies have shown that 1.6 Gy/fraction results in equivalent tumor control with 2.0 Gy/fraction [5], early responding tissues with a similar α/β ratio such as mucosa and dermis should conversely have a comparable extent of damage from and recovery between fractions [16]. Therefore, doses within this range of 1.6–2.0 Gy should not significantly alter the overall effect on acute toxicity on a per-fraction basis. However, with fewer total fractions to normal tissue with sequential boost IMRT, the result would be decreased severity of acute toxicity.

The addition of concurrent chemotherapy to definitive radiation for locally advanced head and neck cancer is well known to increase toxicity compared to radiation alone [17‐19]. Concurrent chemotherapy regimens are variable, but historically these have been cisplatin based or have included 5-FU. In our cohort, radiation was given with concurrent weekly carboplatin/paclitaxel and many also received weekly induction chemotherapy with the same agents. While cisplatin and carboplatin/paclitaxel have not been directly compared in a randomized fashion, this combination is often utilized as primary induction or concurrent chemotherapy or for patients unable to tolerate cisplatin, and our group and others have reported comparable treatment outcomes and acceptable toxicities [13, 20‐22]. Furthermore, rates of high grade toxicity with concurrent carboplatin/paclitaxel seen in our SIB cohort are comparable with rates reported with cisplatin or 5-FU-containing regimens [13, 15, 17, 19]. Nevertheless, whether the differences seen in acute toxicity between SIB and sequential boost in our study would apply to patients receiving traditional cisplatin-based chemotherapy is not entirely clear and would require further evaluation.

A majority of patients in both cohorts also received induction chemotherapy prior to chemoradiation. Induction carboplatin/paclitaxel was typically offered to individuals with bulky tumors or lymphadenopathy and/or lower cervical lymphadenopathy, and was previously shown to be well-tolerated with 15% or less grade 3 and 0–1.5% grade 4 hematologic toxicity [13]. The potential impact of induction chemotherapy on subsequent toxicity during chemoradiation is difficult to ascertain in our study given the small fraction of patients who did not receive induction. However, the PARADIGM trial reported no difference in rates of mucositis, pain, xerostomia, or feeding tube dependency between those who received induction chemotherapy and those who did not [23]. Additionally, a recent meta-analysis confirmed that induction chemotherapy had no impact on rates of non-hematologic toxicity during the chemoradiation portion of therapy as compared to chemoradiation alone [24]. Furthermore, we found no significant difference in the utilization of induction chemotherapy between our SIB and sequential boost cohorts. This suggesting that induction chemotherapy is unlikely to contribute significantly to rates of acute toxicity reported in our study or the difference observed between cohorts.

As with chemotherapy regimens, there is often some degree of variability in radiation dose and fractionation used for definitive treatment of LAHNC. In our cohort, gross disease was treated to 69.3 Gy in 2.1 Gy per fraction. This is slightly hypofractionated when compared to the traditional 70 Gy in 2 Gy per fraction, however our dose and fractionation is similar to that used for concurrent treatment in nasopharyngeal carcinoma cooperative studies and results in comparable BED₃ and BED₁₀ to 70 Gy in 2 Gy fractions [15]. Additionally, dose to the prophylactic volumes is within the standard range for both the SIB and sequential boost groups. For these reasons, we feel that our reported results are applicable to other common LAHNC radiation doses and fractionations.

Most prior studies comparing SIB and sequential boost IMRT for LAHNC examined the differences from primarily a dosimetric standpoint. To our knowledge, our study is the largest to date exploring the clinical consequences of these distinct treatment approaches and the first to examine this difference among the most common head and neck subtypes. Recently, Songthong et al. reported on their phase II/III trial comparing SIB and sequential boost IMRT in patients with nasopharyngeal carcinoma and found no significant difference in short-term treatment outcomes or acute toxicities [11]. While the prospective nature of this study circumvents some of the limitations of ours and appears to contradict our findings with regard to toxicity, a number of factors call into question the strength of their conclusions and broader applicability. First, they utilized a different dose/fractionation scheme to gross disease in the SIB and sequential boost arm (2.12 Gy versus 2.0 Gy per fraction). Second, the study was limited to patients with nasopharyngeal carcinoma, which is significantly less common than oropharyngeal and laryngeal carcinoma. Third and most importantly, Songthong et al. report very low rates of grade ≥3 acute toxicity in either group when compared to historical controls (2–15% vs approximately 70%) [11, 15, 17, 19]. Furthermore, the study was powered to detect a generous 20% difference in acute toxicity rates yet did not reach the number of enrolled patients to achieve that (122 versus target of 218) [11]. With unusually low rates of high grade toxicity at less than 20% and effectively being powered to detect greater than a 20% difference, the lack of a significant difference may be due to the fact that this study was underpowered. In contrast, we report on a larger cohort with rates of grade ≥3 acute toxicity more in keeping with historical controls and were able to see a significant difference in various acute toxicities.

In general, SIB and sequential boost IMRT offer directly competing advantages and disadvantages with regard to radiobiology and treatment planning. Relative to SIB, sequential boost IMRT allows for consistent dose-per-fraction between all treatment volumes and throughout the course of treatment. This addresses the theoretical concern of under-dosing target volumes with 1.6–1.7 Gy fractions, particularly in the latter weeks of treatment when accelerated repopulation of tumor clonogens is prevalent [25, 26]. Additionally, sequential plans can better account for changes in body habitus (weight loss), edema, and tumor shrinkage when a new CT is obtained for boost planning. On the other hand, sequential boost IMRT is more time consuming and involves the summation of 2 or more treatment plans, which can theoretically result in more potential uncertainty in true dose distribution (particularly if based on separate CT simulations). Centers with high volume and/or limited or inexperienced staff find this more challenging compared to a single SIB plan. Continuous improvements in treatment algorithms and decreased time required for planning have improved the overall throughput for sequential boost IMRT, though these relative challenges still remain. Nevertheless, our data suggest that despite their differences, these two techniques are comparable with regard to treatment outcome.

As a retrospective analysis on two cohorts of patients treated at distinct periods of time, this study is limited by the lack of a prospective comparison of the two methods of treatment planning. These differences are partly mitigated by the use of consecutive patients, similar chemotherapy regimens, similar dose-volume constraints for IMRT, and prospective grading of toxicity by a single, experienced radiation oncologist at the time of treatment. The latter greatly minimizes the potential for intra- and inter-observer variability in these more subjective values. Furthermore, while the SIB and sequential boost patients were treated during non-overlapping intervals of time between 2003 and 2012 with the SIB group treated earlier, any differences in treatment planning, evaluation, and delivery were minimized over that span of time due to a number of factors. Specifically, contouring practices, treatment machines, dose calculation algorithms, and treatment planning software were consistent during this time. Still, there may be less quantifiable changes between these time periods that cannot be adequately accounted for. Additionally, changes in acute toxicity were not associated with gastrostomy placement or duration, weight loss, or changes in late toxicity in our cohort. Therefore, all these potential differences will need to be accounted for or confirmed in a formalized, prospective manner.

Concurrent chemoradiation with sequential boost IMRT for LAHNC is well-tolerated and results in treatment outcomes comparable to simultaneous integrated boost. Our institutional experience also suggests that sequential boost reduces high-grade acute toxicity, and these data warrant further evaluation in a prospective study. In this setting of a potential reduction in acute toxicity, sequential boost IMRT is especially appealing as an alternative to SIB as image-guidance usage increases, particularly on-board cone beam CT. This could potentially reduce toxicity even further through real-time, clinically-guided modification of target volumes. Coupling image-guided assessment of tumor response with sequential field size reductions may allow for truly adaptive radiotherapy in a way not possible or practical with simultaneous integrated boost.