Rectal cancer irradiation. Long course, short course or something else?

Preoperative radiotherapy has repeatedly been shown to reduce the risk of a local failure in a primary rectal cancer considered to be resectable. In a locally advanced rectal cancer with overgrowth to surrounding non-resectable structures (the more advanced T4s), preoperative radiotherapy, presently together with chemotherapy [1], may allow radical (R0) resection in a substantial proportion of the cases. The ability to decrease local failures in the resectable cases is seen also if surgery is optimized, according to the total mesorectal excision (TME) principles [2]. Postoperative radiotherapy, even if combined with chemotherapy has less effects and is also more toxic than preoperative radiotherapy, as shown in several randomised trials [3–6].

The preoperative hypofractionated, short course 5×5 Gy radiation schedule and immediate surgery has ever since it was originally explored in the late 1970s, been an attractive alternative to a conventional long course schedule using 1.8–2 Gy per fraction to a dose of about 45–50 Gy. Actually, more patients have probably been included in randomised trials where 5×5 Gy has been one treatment group than in trials exploring conventionally fractionated schedules. The 5×5 Gy schedule reduces local failure rates and improves survival with the previous type of surgery (blunt dissection) [7–9] and reduces local failure rates with TME [2]. The incremental gain in local failure reduction together with TME (about 6%-units) is likely too small to show-up as a survival gain even in a trial including 1 800 patients [10]. Recently, preoperative short course 5×5 Gy was as effective as preoperative long course radiochemotherapy (50 Gy with 5FU) in one trial (local failure rate 11% vs 16%, p = 0.23, identical survival) [11] and superior to postoperative radiochemotherapy (45 Gy + 5FU) given to patients at high risk for a local failure (local recurrence at 5 years 5% vs 17%, p < 0.0001; disease-free survival 75% vs 67%, p = 0.03; overall survival 72% vs 62%, p = 0.07) in another trial [5]. Acute toxicity was in both trials much less using 5×5 Gy, and late toxicity did not differ, however, only with limited follow-up.

Hypofractionation, i.e. giving the radiotherapy in fractions above 2 Gy has the advantage of shortening treatment times and reducing the burden to radiotherapy departments. Hypofractionation has, however, also the disadvantage of diminishing the therapeutic index with a particular risk of increased late toxicity. Actually, experience has repeatedly learned that many hypofractionated schedules have resulted in unacceptable late toxicity rates and there is a wide spread, and sound scepticism against any type of hypofractionation. The most recent trends in increased use of hypofractionated radiotherapy in for example breast [12] and prostate cancer [13] have raised genuine concerns of ‘do we have to learn the lesson once more’.

These concerns are naturally present also for the rectal cancer 5×5 Gy schedule not only among those who for theoretical or practical reasons dislike hypofractionation but also among those running the trials. Actually, much more knowledge about late adverse effects are known from the 5×5 Gy trials than for any of the conventional radio(chemo)therapy schedules. We thus reasonably well know the scale of morbidity that can be seen after 5×5 Gy with a follow-up to about 15 years [3], [14–20]. For most rectal cancer patients, where median age at diagnosis is above 70 years, a 15 year follow-up is sufficient, but even longer follow-up is necessary since some patients may live for 20–30 years or more after their diagnosis. Late effects from radiotherapy can occur late. The possibilities to deliver radiotherapy more accurately continuously improves, so it can be expected that the adverse effects seen in the trials from the treatments delivered during the 1980s and 1990s will be less in the future [21]. All effective treatments have a price. In the light of the present knowledge of acute and late effects from 5×5 Gy, not to be detailed here (see [3], [14–16], [18–20], it is clear that preoperative radiotherapy is routinely indicated for many patients with rectal cancer, but not for others with a lower risk of failing locally. Which groups of patients who could be exempted from preoperative radiotherapy is not precisely known and must be continuously discussed [22]. To rely on postoperative radiochemotherapy is an attractive option to many, since it excludes many patients from an unnecessary preoperative treatment, but was clearly inferior in the MRC-CR07 trial [5].

The concerns for the late adverse effects of the high 5 Gy fractions, even if ‘only’ 5 fractions are given resulting in a comparably low total dose of 25 Gy, have resulted in attempts to modify the schedule, at the same time keeping the total treatment time short (one or two weeks). Then multiple smaller daily fractions are necessary. In an experimental study, the tumour cell kill was similar using 5×5 Gy and 10×3 Gy (2 daily fractions 6 hours apart) in 5 days, and superior to 10×3 Gy in 10 days [23]. Two of the clinical trials are published in this issue of Acta Oncologica [24], [25]. These studies build upon the limited knowledge of the relations between fraction size, total dose and treatment time on the one hand and acute (tumour and normal tissue) and late toxicity on the other, as expressed in the so called ‘biological models’. These are generally based upon the linear quadratic (LQ)-model.

Our own concerns about the 5 Gy fractions lead to a phase I trial in Uppsala between 1986–1987 (Glimelius and Påhlman, unpublished). Successive cohorts of patients with rectal cancer, routinely scheduled for 5×5 Gy, could participate in a trial delivering increasing doses of radiotherapy twice daily with a minimal interval of 6 hours for 2 weeks, starting at a fraction dose of 1.5 Gy (total dose 30 Gy). Three patients were treated at each level, and if no toxicity was seen, the dose was increased by 0.05 Gy per fraction. If one of the patients had some toxicity, three or more patients were treated at that level, whereas if two patients had toxicity, this dose was too high. At a fraction size of 1.6 Gy, one of six patients had visible oedema and acute reactions in the pelvis cavity during surgery and at 1.65 Gy, this was seen in two of three patients. In addition, one patient got neurophatic pain after about 12 fractions in a way that has been seen after 5×5 Gy [26]. The trial was terminated since the target goal with a treatment that resulted in supposed equivalent tumour cell kill to 5×5 Gy was not reached. The extra resources required (20 rather than 5 fractions) also influenced the decision.

Brooks et al. [24] from Mount Vernon Cancer Centre, UK treated 20 patients with clinical stage T3 rectal cancer with a continuous hyperfractionated, accelerated radiation therapy (CHART). The patients received 25 Gy in 15 fractions of 1.67 Gy in 5 days, i.e. the same total dose, daily dose and number of treatment days as in the 5×5 Gy schedule. However, since the daily dose of 5 Gy was split into three fractions, 6 hours apart, the predicted late toxicity will be much reduced (from BED (α/β = 3 Gy) 66.7 Gy using 5×5 Gy to BED 39.0 Gy, Table I). This is a substantial decrease in the predicted risk of late toxicity that could motivate the increased work load to the radiotherapy department. However, at the same time, the predicted tumour control probability (TCP) (α/β = 10 Gy) is reduced, although comparably less (from 37.5 Gy to 29.2 Gy). The 5×5 Gy schedule has in the trials reduced the local recurrence rate by 50–65% [2], [7], [9]. In a subgroup analysis of updated results from the TME trial [2], no statistically significant decrease in local recurrence rate was seen in the group of patients with positive circumferential margin (crm + ) (from 23.3% to 15.5%, p = 0.16, a relative reduction (RR) of 33%) [27]. In patients with negative crm (crm-), a more marked reduction was seen (from 9.1% to 3.7%, p = 0.001, RR 59%). A smaller relative reduction in the crm+ patients with likely more tumour cells to kill than in those with crm- is logical, and in line with other data [28]. The decrease in predicted TCP is thus of concern.

Table I.  Biological equivalent doses

	Tumour control/acute normal tissue complication probability		Late normal tissue complication probability
	BED (Gy) (α/β = 10 Gy)		BED (Gy) (α/β = 3 Gy)
Regimen	No time correction	With time correction
25 Gy/5 fr/5 days (d = 5 Gy)	37.5	37.5	66.7
50 Gy/25 fr/33 days (d = 2 Gy)	60.0	44.4	83.4
25 Gy/15 fr/5 days (d = 1.67 Gy)	29.2	29.2	39.0
25 Gy/10 fr/5 days (d = 2.5 Gy)	31.3	31.3	45.8
41.6 Gy/26 fr/17 days (d = 1.6 Gy)	48.3	42.3	63.8
42 Gy/28 fr/18 days (d = 1.5 Gy)	48.3	41.7	63.0
33 Gy/20 fr/12 days (d = 1.65 Gy)	38.4	35.4	51.2

Equation 1. Linear quadratic based isoeffect, basic formula without time correction, BED = nd (1 + d/α/β), where n = number of fractions, d = dose (Gy) per fraction, α/β = the LQ quotient.

Equation 2. Time-corrected LQ-formula, BED = nd (1 + d/α/β)-γ/α (T − T_k), where γ/α = repair rate (set to 0.6 Gy/day), T = overall treatment time and T_k=proliferation delay (set to 7 days, or maximally T) [9].

No compensation was made for multiple fractions per day, due to incomplete repair, since this did not have any practical importance using a 6 hour inter fraction interval.

The CHART study [24] is too small to evaluate antitumour activity, although 3 year locoregional control was good (95%). Acute toxicity was, as expected low, but one patient had to stop therapy due to nausea and vomiting (virtually never seen after 5×5 Gy [29], [30], and two (10%) died postoperatively. This figure is much higher than otherwise seen (2–4%), but could be caused by chance.

In yet another study, 25 Gy was given during 5 days with a daily dose of 5 Gy, however, split into two doses of 2.5 Gy [31]. This schedule, although still hypofractionated, will also reduce late normal tissue complication probability (NTCP) compared to 5×5 Gy, but also TCP, Table I. This group from Vienna, Austria treated 184 patients, again with clinical stage T3 (about one third had pathological stage I) between 1994 and 2004. Treatment was tolerable also during follow-up, and tumour control was excellent (2.1% local failures at 4 years). However, even if the results are very favourable, the true benefit of this schedule, compared to 5×5 Gy, can not be judged because of the uncontrolled design.

A Swiss group [25], [32] has between 1993 and 2002 treated 279 patients (250 assessable) preoperatively with 41.6 Gy in 2.5 weeks at two daily fractions of 1.6 Gy. This hyperfractionated accelerated schedule (HART) is similar to the one we tried in the mid 1980s. Similarly, in a Polish study [33], 62 patients were treated with HART (42 Gy during 2.5 weeks at two daily fractions of 1.5 Gy). These schedules result in the highest predicted TCP of all the “short-course” schedules, and a predicted late NTCP in the same range as that from 5×5 Gy (Table I). The acute normal tissue reactions appeared in both studies tolerable, although some subjective toxicity was seen in many patients. In the Polish study [33] it is noteworthy that 50% (11/22) of the patients who underwent abdomino-peritoneal amputation had prolonged perineal wound healing (>2 months). Treatment outcome with a local failure rate of 6–8% after 3–5 years is favourable, considering the stage distribution (clinical stage T3-4 or T1-2N+, pathological stage I (or 0) seen in 25–27%). When the Swiss group analyzed outcome depending upon the time interval between the end of the radiotherapy and surgery, those with a gap duration of 5 days or more had significantly better overall and disease-free survival [25]. There was no difference in local failure rates, but since few local failures were seen, this is difficult to evaluate. The authors give several possible explanations for this unexpected finding. Since a longer gap will result in more down-sizing and down-staging of the primary tumour, the surgery could be facilitated, potentially resulting in fewer local failures. This has, however, never been proven. Experience has shown that a long delay before surgery of 5–6 weeks after 5×5 Gy can result in down-sizing, and previously inextirpable tumours can be resected (Glimelius B, Påhlman L, Holm T, unpublished observations) [34]. A randomised phase III study (Stockholm III) is presently ongoing, comparing 5×5 Gy with immediate or delayed surgery or with 25×2 Gy with delayed surgery. About 330 patients have been included (Cedermark B, pers. comm.).

In spite of the theoretical concerns and many predictions speaking against the 5×5 Gy schedule (the map), the trials and clinical experience (reality) have shown that it works well. This discrepancy between the map and reality of course leads to the question of what is wrong. This has among others been expressed by late Professor Juliana Denekamp, a prominent radiobiologist, in the symposium to her honour [35], [36]. Few single treatments in oncology have been the subject of so many large scale trials, having the ambition to carefully also report not only long-term gains [17] but also long-term adverse effects. Mistakes have been made in the design of the trials with too many simplifications in important technical details, resulting in too much both acute and late toxicity. This was a catastrophe for those severely suffering from the mistakes but also for all those indirectly suffering of not getting a valuable treatment because of an unfounded fear for untoward effects, even if the treatment was properly given. The short course preoperative 5×5 Gy schedule is, according to present knowledge, at least as effective as the most efficient long course therapy known today, about 50 Gy with 5FU-based chemotherapy. Whether this will remain so in the future with better sensitizing agents, particularly the biologicals [37], can only be speculated upon. Since it is not immediately apparent that 5×5 Gy safely can be combined with any drugs, attempts to modify or develop new schedules like the ones now reported, are urgently needed.