Conventional fractionation should not be the standard of care for T2 glottic cancer

Between January 1999 and December 2005, 112 patients were treated for T2N0 glottic squamous cell carcinoma. There were 98 (88%) males and median age was 64 (range, 38–85 years). Seventy-six patients (68%) were staged as T2a, and 36 (32%) had T2b disease, as shown in Table 2.

Although 5-year LC in smokers was worse than that for non-smokers, this was not statistically significant (78% vs 87%, p = 0.42). Younger patients aged under 65 did significantly worse than older patients, with inferior 5-year LC (73% vs 94%, p = 0.003).

The median follow up was 5.8 years (range 0.1–9.6 years). All but four cases were followed up for at least 3 years, or had disease recurrence before that point. Nineteen (17%) patients developed recurrence, of these 11 (9%) were within the radiotherapy field, 4 (4%) were nodal recurrence alone, and 4 (4%) were both in field and nodal recurrences. There were 18 (16%) patients who developed second malignancies in the lung, kidney or bladder.

The 5-year OS was 67%, 5-year LC was 82%, and 5-year DSS was 90%. The respective 5-year LC rate for T2a and T2b disease was 88.8 and 70.8% (p = 0.032). Age less than 65 years was associated with poorer outcome, 5-year LC of 73.2%, compared with 93.9% in those aged 65 or over (p = 0.003), as shown in Table 2.

There was a low rate of severe late morbidity (defined as requiring surgical intervention); 2 (1.8%) patients who required a tracheostomy or total laryngectomy. Twelve (10.7%) patients had a total laryngectomy for disease recurrence.

For T2N0 glottic larynx cancer, a hypofractionated schedule of 52.5 Gy in 3.28 Gy per fraction, resulted in 5-year loco-regional control rate of 82%. This is comparable with other hypofractionated treatment regimens, demonstrating local control rates of 76–82% [9, 10, 15]. However, the current international standard dose fractionation for head and neck cancer of 70 Gy, in 2 Gy per fraction (Gy/#), delivered over 7 weeks is associated with a 5-year LC rate of 68–75% [7, 16‐18]. This raises the question as to whether hypofractionated schedules for early glottic larynx cancer should be adopted as a new standard. Despite excellent outcomes for T2a tumours, results were more disappointing for T2b tumours, with only 71% 5-year LC, and therefore alternative treatment strategies using either surgery or chemoradiotherapy, should be considered, and are discussed later.

In our cohort there was no significant difference in local control dependent on haemoglobin levels, although this has been shown to have a prognostic impact on local control in other series [5]. A recent meta-analysis also found low pre-treatment haemoglobin to be a major influential factor for radiation failure in early glottic cancers (relative risk 0.891, p < 0.001) [19]. Another unexpected result was a statistically significant inferior local control in younger patients aged under 65. Generally it would be expected to see worse outcomes in an older cohort, but there was no multivariate analysis on our subgroups to interpret this result further.

Outcomes from radiotherapy for early glottic larynx cancer have improved over time [6, 8, 9], likely due to multiple factors including; improved staging accuracy with introduction of routine pre-treatment CT scanning, all patients being discussed at an multidisciplinary team meeting, and national cancer guidelines for head and neck [20, 21]. Glottic tumours with inner cortex thyroid cartilage invasion and/or cord fixation and/or invasion of the paraglottic space are staged as T3. While flexible nasendoscopy will demonstrate a fixed hemi-larynx, thyroid cartilage invasion and paraglottic space invasion are best assessed on CT. Several studies have demonstrated that endoscopic assessment frequently understages laryngeal cancer and demonstrated the importance of CT imaging in providing accurate staging information [22, 23]. As patients are now staged more accurately, due to the advances in imaging technology, this has potentially resulted in more patients being upstaged to T3 and an apparent improved outcome for T2 tumours (Will Rogers effect) [23].

From a radiobiological perspective, the main factors influencing local control are total dose, dose per fraction and overall treatment time. The DAHANCA6 trial compared six fractions per week with five fractions per week, using 2 Gy/#, showing a reduced risk of recurrence with accelerated regimes for T1–T2 glottic cancers (HR 0.6, CI 0.41–0.89) [24]. This study also found benefit from acceleration in well and moderately differentiated tumours, suggesting that accelerated repopulation requires cells to have adequate functional behaviour to respond to the radiation trauma, and these mechanisms are more pronounced in differentiated cells [24]. Our study was limited by lacking histological differentiation for our cohort, which may significantly affect outcomes from our accelerated fractionation regime.

To compare local control outcomes for different schedules, BED was calculated for different dose fractionation regimens, using the following equation [25]:

Calculations included time lag and time factor, to account for the effect of repopulation. Accepted values for time factor range from 0.5 to 1 [26, 27], while time lag is usually considered to be between 21 and 28 days [25, 27]. Assuming fraction size to be more influential than overall treatment time, the initial value chosen for time factor was 0.6, with a time lag of 28 days. Two regimens that result in similar local control rates of 70% are 70Gy in 35# over 7 weeks, and 50Gy in 20# over 4 weeks. The BED for the tumour (BED10) was therefore calculated for these regimens, using the above equation, (using α/β of 10):

The values for these regimens do not correlate with their similar local control rates, and show a 10Gy difference in BED. To place more emphasis on overall treatment time, the value for time lag was changed from 28 to 21 days, and BED10 recalculated:

These values remain different by 10Gy, suggesting that time factor is more influential, hence this was increased from 0.6 to 1.0, and BED10 was recalculated using an intermediate value of 25 days for time lag.

These BED values are much more similar, in accordance with the similar local control rates for these regimens. Literature was reviewed to determine 5-year LC outcomes for T2 glottic cancers, where radiotherapy doses were quoted to allow BED to be calculated (Table 3). These BED values were plotted against local control rates to give the dose response graph (Fig. 1).

There are weaknesses in radiobiological modelling for tumour and normal tissue parameters using a linear quadratic formula including heterogeneity in patient, tumour and treatment characteristics. Even when homogeneous data are used to characterise tumour control (Table 3), assumptions are made for α/β ratio, duration of lag phase and time factor [11]. Accepting these weaknesses, the dose response gradient for T2 glottic cancer shows an absolute increase in local control of nearly 12%, with a 10Gy increase in BED. Previously reported data for T3 disease has shown an increase in local control of approximately 15% for a 10Gy increase in BED, albeit calculated using different time parameters [28].

The other consideration is that while improved local control is an attractive prospect, this should not be at the expense of increased serious late toxicities. However, shorter hypofractionated regimens for early glottic larynx cancer, which use small treatment volumes, have a low rate of severe late effects, approximately 1.1–1.8% [10, 15], which is comparable to a conventionally fractionated schedules of 70 Gy in 2 Gy/# (1.3–2.6%) [7, 18]. Studies of T1 glottic cancers using 3.125 Gy/# have shown no severe acute complications following treatment, which suggests hypofractionated regimes are safe and well tolerated [29, 30]. Dinshaw et al. reported persisting radiation oedema in 28% of patients with T2 glottic cancers treated with 2–3.3 Gy/#, however there was no correlation between toxicity and increasing fraction size [31]. Our toxicity data is limited by being retrospective, and only recording severe grade 4 toxicity requiring surgery, so a recommendation for further studies would be to record all grades of toxicity prospectively.

Using further radiobiological modelling, where α/β = 3 for late responding normal tissues, the biologically effective dose (BED3) was calculated for both conventional and hypofractionated regimens, to compare the impact on these late effects (Table 3). BED3 was higher for the conventional dose fractionation schedule, and lower for hypofractionated treatments (Fig. 2), supporting the finding that late effects were not increased with these hypofractionated schedules.

As suggested by previous studies, T2b compared with T2a tumours had worse local control rates [4, 32]. We suggest the subdivision of T2 glottic tumours into T2a (normal vocal cord mobility) and T2b (impaired mobility) should be included in the TNM classification. Synchronous platinum based chemotherapy with radical radiotherapy is now standard of care for T3 glottic cancers and has been shown to improve outcomes in T2 disease [33‐35]. A retrospective review by Bhateja et al. showed a trend towards inferior local control for T2b cancers treated with radiotherapy alone, compared to those with more advanced disease treated with chemoradiotherapy [33]. A limitation of this study was a lack of toxicity and quality of life data, which is important when balancing risks and benefits of intensifying treatment. Akimoto et al. investigated 63 patients with T2N0 disease, and found that chemoradiotherapy improved 5-year OS and 5-year disease-free survival compared to radiotherapy alone (96 and 89% vs. 87 and 68%, p < 0.05) [34]. In a multicentre review, Hirasawa et al. found an improvement in 5-year LC with chemoradiotherapy for T2 glottic cancers compared to those treated with radiotherapy alone (80.7% vs 64.4%, p = 0.149) [35]. Given the poorer outcomes associated with T2b disease, it may be worth considering concurrent chemotherapy in these patients, where the benefits may outweigh the potential increase in toxicity.

Systematic reviews of early glottic cancers have found similar outcomes between radiotherapy and surgery for patients with T2 glottic cancers [2, 36]. The average 5-year LC was 76% for radiotherapy compared to 77% for transoral laser microsurgery (TLM), however there were no randomised control trials identified for inclusion in this review. Similar outcomes for T2 glottic cancers treated by TLM were reported by Ansarin et al., with 75% 5-year disease-free survival [37]. Grant el al. achieved 93% 5-year LC for T2 glottic cancers after TLM, suggesting this is a safe and effective treatment option [38].

Voice quality following surgery is generally felt to be inferior to radiotherapy, however a review by Speilmann et al. found comparable vocal and quality of life outcomes for radiotherapy and TLM for early glottic cancers [39]. The ENT-UK Head and Neck Group recommended that TLM is offered to all patients with early glottic tumours up to T2a, as a standard of care [40]. Given the poorer outcomes for T2b glottic tumours, alternative management options including surgery should be considered for this subgroup of cancers.