Keywords

Introduction

Human papillomavirus (HPV)—mediated oropharyngeal cancer (OPC) has been rapidly increasing in incidence over the past few decades [1, 2]. As HPV-positive OPC often affects younger people and demonstrates high survival rates, this means that patients will usually live with the morbidity of treatment for decades to come. As a result, the concept of de-escalation of treatment has found widespread acceptance as a possible solution to this problem.

Why De-escalation?

The seminal RTOG 0129 trial showed that HPV-positive OPC demonstrates considerably better overall survival than HPV-negative disease. It also described three risk groups, depending on HPV and smoking status. The lowest-risk patients demonstrated especially high rates of cure with over 90% three-year survival [3]. These patients, who are often younger than the traditional HPV-negative head and neck cancer patients, may often live for many decades with the long-term toxicities of their treatment. Importantly, it is widely acknowledged that treatment with cisplatin increases the number of acute serious toxicities by a factor of two, compared to radiotherapy alone [4]. It also demonstrates long-term and lasting toxicities such as swallowing dysfunction [4, 5].

Therefore, the concept of de-escalation has gathered momentum over the last decade. This concept espouses the reduction of treatment doses, or the use of alternative treatments to reduce toxicity, whilst maintaining the excellent survival rates demonstrated with standard chemoradiotherapy. Treatment de-escalation can take place in many forms; including the substitution of cisplatin for other potentially less toxic agents, for example cetuximab; the introduction of induction instead of concomitant chemotherapy with radiotherapy; the reduction of the dose of radiotherapy; the elimination of chemotherapy altogether; and the use of single modality treatment, either surgery or radiotherapy alone, instead of combination therapy. As a result, many studies were initiated, now almost a decade ago. The results of the first studies to report have raised some interesting, and indeed at times disturbing, conclusions which necessitate a re-assessment and reconsideration of our strategies.

It is important at this point to ask whether the concept of de-escalation is supported by patients. This question was elegantly addressed by the work by Brotherston et al. [6] who interviewed patients with OPC who had been treated with chemoradiotherapy about their preferences for de-escalation. Ninety-nine percent of the respondents favoured de-escalation of treatment if that did not result in a difference in overall survival. However, if it reduced survival then only 69% of the patients interviewed would support any form of de-escalation, and only up to a detriment of no more than 5% in survival rates. Eighty-one percent would prefer to avoid chemotherapy rather than radiotherapy. Therefore, from that study we surmise that patients are supportive of de-escalation, but only if it has a minimal effect on survival and efficacy of treatment [6].

Results of Recent De-escalation Trials

Several de-escalation studies have been reported in the last five years. The first to report were the De-ESCALaTE [7] and RTOG 1016 [8] randomised controlled trials. Later the TROG 12.01 study also reported [9]. These compared concurrent cisplatin with cetuximab, in conjunction with radiotherapy. Cetuximab had been reported in the Bonner trial [10] to improve survival when added to radiotherapy, with relatively little reported additional toxicity. Cetuximab was therefore widely considered to be less toxic than cisplatin. However, the De-ESCALaTE [7], RTOG 1016 [8] and TROG 12.01 [9] studies all showed similar and surprising results. These studies demonstrated a significant additional benefit from cisplatin, compared with cetuximab, both loco-regional control and overall survival. The RTOG 1016 trial [8] demonstrated an estimated overall survival at five years of 84.6% (95% CI, 80.6–88.6%) for cisplatin compared to 77.9% (95% CI, 73.4–82.5%) for cetuximab. Similarly, the De-ESCALaTE study [7] showed overall survival at two years of 97.5% for cisplatin, compared to 89.4% for cetuximab, demonstrating an adjusted hazards ratio of 5.0 (95% CI 1.7–14.7; log-rank P = 0.0012). This difference in survival was seen even in the lowest risk HPV-positive OPC, that is when excluding T4 and N3, in the De-ESCALaTE study. In this latter group, there was a two-year overall survival difference of 5.2% between the two groups with a hazards ratio of 4.3 (95% CI, 0.9–19.8; log-rank P = 0.0431), in favour of cisplatin. The TROG12.01 study [9] showed a significant difference in 3-year failure-free survival rates, which were 93% (95% CI, 86–97%) in the cisplatin arm and 80% (95% CI, 70%-87%) in the cetuximab arm (hazard ratio = 3.0 [95% CI, 1.2–7.7]); P = 0.015.

Importantly, in all the studies there was no significant difference in toxicity between the two arms. For example, in the De-escalate study [7] the incidence of all grade or severe (grade 3–5) toxicity, both in the acute and late phases between the two groups. The mean number of overall severe (grade 3–5) toxicity events per patient was 4·8 [95% CI 4·2–5·4] with cisplatin vs 4·8 [4·2–5·4] with cetuximab; p = 0·98). The mean number of late severe toxicity events was 0·41 (0·29–0·54) with cisplatin and 4·82 (4·22–5·43) with cetuximab. The types of toxicity differed between the arms, as would be expected.

The next randomised study to report was NRG HN002 [11]. This phase II study randomised patients into accelerated intensity modulated RT (IMRT), at a dose of 60 Gy in 5 weeks, with weekly cisplatin (40 mg/m2/week), against IMRT alone. Both of the arms were experimental, because of the reduced radiotherapy dose, and there was no comparison with standard chemoradiotherapy regimens. The hypothesis was that an arm would be taken forward into a larger phase III study if it achieved a historical control rate of progression-free survival (PFS) rate at two years of equal or more than 85%. The arm also had to show a mean one-year MD Anderson Dysphagia Inventory (MDADI) composite score of more than 60, demonstrating acceptable swallowing toxicity. The concomitant cisplatin and IMRT arm reached the pre-specified criterion, with a two-year PFS of 90.5%. However, the IMRT alone arm demonstrated a two-year PFS of 87.6% (P = 0.23) and therefore did not meet the pre-specified endpoint. It again demonstrated the importance of the addition of cisplatin even in the lowest risk p16 positive patients [11].

More recently, the ECOG 3311 phase II study [12] was published. This randomised study looked at the role of dose de-escalation of adjuvant radiotherapy, in conjunction with cisplatin. Patients with HPV-positive OPC received transoral surgery. If, on post-operative histology, they were low risk, they were not given any adjuvant treatment, and if they were high risk, they were recommended adjuvant chemoradiotherapy. If they were determined to be at intermediate risk of recurrence and were recommended adjuvant radiotherapy alone, they were randomised to receive either 50 or 60 Gy post-operative radiotherapy. The results of the two randomisation arms were very similar, with the lower dose radiotherapy 50 Gy arm showing a two-year PFS of 94.9% (90% CI, 91.3–98.6) and for the standard 60 Gy arm 96.0% (90% CI = 92.8–99.3). This study showed the possibility of reducing the dose for post-operative patients who did not require cisplatin and is the first evidence of the possibility of de-escalation [12]. The Pathos trial [13] is looking to demonstrate the same as the ECOG 33–11 trial, but also extend de-escalation to the high-risk group of patients by eliminating cisplatin from the experimental arms. This study has progressed from the phase II stage, and is now recruiting in phase III.

There have been other de-escalation studies, but these have all been phase II, non-randomised studies and are therefore difficult to draw conclusions from. Some have shown interesting, and possibly promising results, but they should all be considered hypothesis generating, and all require validation in randomised phase III studies.

Lessons Learnt from De-escalation Studies

The first lesson that was learnt from these de-escalation studies was that concomitant chemoradiotherapy with cisplatin is a highly effective treatment for patients with HPV-positive OPC. Substitution of cisplatin or its elimination appear to have a significant detrimental effect in several studies. This is the case even in very low risk patients, as demonstrated in the HN002 and the sub analysis of the De-ESCALaTE study [7, 11]. It should therefore be clearly understood and widely appreciated that de-escalation of treatment in patients with HPV-positive OPC can have detrimental effects on patients’ survival, and therefore must be studied in a highly controlled manner.

The second lesson follows on from the first one. Because of the potential for significant detriment, studying de-escalation should be in the context of randomised phase II trials. Running phase III trials from the start, without a preceding phase II study, should be avoided and strongly discouraged. We saw that the De-ESCALaTE study [7] demonstrated a similar result to RTOG 1016 [8]. However, it utilised considerably fewer patients: 334 compared to 844 respectively. This meant that fewer patients came to harm as a result of being part of the study, whilst demonstrating the same effect. This was also demonstrated in both HN002 (which randomised 308 patients) and in the randomisation arms of ECOG 3311 (which randomised 359 patients). Indeed, the TROG 12.01 study [9] showed a significant difference in failure free survival with only 189 patients randomised. Therefore, when it comes to de-escalation in the context of HPV-positive OPC, it would appear that studies randomising approximately 300–350 patients are sufficient to demonstrate significant differences between the two arms. Once the results of the phase II studies are available, then the phase III studies should be started, and not before.

An example of good practice is NRG HN005, which is now taking the results of HN002 into a new phase II study, which is comparing the successful arm of HN002 against the standard of care arm (concomitant cisplatin and radiotherapy), and against a third, new experimental arm of 60 Gy with nivolumab 240 mg in six cycles. This study would then progress to a randomised phase III study, comparing one or two experimental arms against the standard chemoradiotherapy arm. Very recently, a press release by NRG announced that the cisplatin plus 60 Gy arm has been suspended because it has not met its prespecified non-inferiority criteria.

The third lesson is that because concomitant cisplatin and radiotherapy are such an effective treatment for HPV-positive OPC, consideration of other ways of reducing toxicity, as alternatives to de-escalation, should be. These we term ‘harm minimisation techniques’, rather than de-escalation. Such strategies could include the use of proton therapy for selected oropharyngeal cancers instead of IMRT, which is currently being tested in the TORPEDO study. Other studies are looking at reducing the volume of radiotherapy delivered; for example, the INFIELD study is looking at elective volume de-intensification [14]. Similarly, the Canadian CCTG HN11 trial, currently recruiting, is exploring the elimination of radiotherapy to the contralateral side of the neck following a SPECT CT scan, compared to elective bilateral neck radiotherapy, which is the standard practice in many centres. Another study, the phase II AVOID study is examining omitting radiotherapy to the resected primary tumour bed after transoral robotic surgery, if there are negative surgical margins. The mean dose received to the primary site as a result of radiation delivered to the neck is therefore greatly reduced—around 36.9 Gy. These trials have the potential to demonstrate reduced toxicity, whilst maintaining high survival outcomes, and should be supported through recruitment.

Conclusions

In conclusion, de-escalation of HPV-positive OPC can result in patient harm, and must be undertaken in highly-controlled phase II trials. If no detrimental signals are seen, then these treatments can be tested further in phase III trials. Different modalities of harm reduction—what we call ‘harm minimisation’ should also be considered. Recently a framework for de-intensification of treatment for HPV-positive OPC patients has been published by the Head and Neck Cancer International Group, which provides guidelines on how best to achieve safe evaluation of new de-escalation treatments [15].