Intraoperative radiotherapy for locally advanced or locally recurrent rectal cancer: Does it work at all?

Abstract

Background. Intraoperative radiotherapy (IORT) has been given for primary and locally recurrent rectal cancer for 30 years. Still, its effect is not clear.

Material and methods. PubMed and EMBASE search for papers after 1989 on surgical treatment and external beam radiotherapy (EBRT) for primary advanced and locally recurrent rectal cancer, with and without IORT. From each center the most recent paper was generally selected. Survival and local recurrence at five years was tabulated for the total groups and separate R-stages. Also, the technique for IORT, use of EBRT and chemotherapy as well as surgical approach was registered.

Results. In primary cancer 18 papers from 14 centers were tabulated, including one randomized and five internally comparing studies, as well as seven studies without IORT. In locally recurrent cancer 18 papers from 13 centers were tabulated, including four internally comparing studies and also five without IORT. Overall survival (OS) and local recurrence rate (LRR) were higher for primary cancer compared to recurrent cancer. Patients with R0 resections had better outcome than patients with R1 or R2 resections. For primary cancer OS and LR rate of the total groups and R0 stages was not influenced by IORT. An effect on R1/R2 stages cannot be excluded. The only randomized study (primary cancer) did not show any effect of IORT.

Conclusion. IORT does not convincingly improve OS and LR rate for primary and locally recurrent rectal cancer. If there is an effect of IORT, it is small and cannot be shown outside randomized studies analyzing the separate R stages.

The frequency of local recurrence (LR) has been extensively reduced after the introduction of total mesorectal excision (TME) surgery [1,2] for rectal cancer. However, still about 10% of the patients experience a local recurrence. For locally advanced primary cancer LR rate may be higher [3]. Many of these patients have LR only and no distant metastases at the time of death, and might therefore possibly be cured if the recurrence could be adequately treated [4].

Preoperative external beam radiotherapy (EBRT), in recent years often combined with sensitizing chemotherapy, has reduced the frequency of local recurrences in locally advanced rectal cancer [5].

Intraoperative radiotherapy (IORT) in combination with EBRT was introduced at the Massachusetts General Hospital and the Mayo Clinic for the treatment of locally advanced and recurrent rectal cancer. Compared to historical controls, patients having IORT were reported to have higher overall survival (OS) and lower LR rate [6,7]. These results were supported by some other studies [8–11], although more recent studies have shown conflicting data [12–15]. Randomized studies have been requested since 1986 but are still virtually lacking [16]. Only one recent study is available on primary rectal cancer combined with external radiotherapy (EBRT) [17] and one without EBRT [18]. These studies showed no effect of IORT on survival or LR rate.

Using IORT the radiosensitive bowel and bladder can be excluded from the radiation field. A higher dose to the tumor bed can therefore be applied. It has been suggested that a single large dose of radiation is equivalent to 2–3 times the same dose when given as 2 Gy fractions [19]. Others have argued that by applying only 1 fraction of radiotherapy a number of the tumor cells will be in a relatively radioresistant state (G0), and therefore will not be susceptible to therapy.

IORT can be given by three different techniques: Electrons (IOERT), high dose rate (HDR) brachytherapy and low dose rate (LDR) brachytherapy using iodine 125 seeds.

IOERT is most often delivered by means of an accelerator during a few minutes. The energy of electrons can be varied according to the depth of tissue to be treated. The radiation dose applied is usually enhanced with increasing tumor burden. Thus, patients with R0- resections are typically given 10–12 Gy, R1-resections 12–15 Gy and R2- or non-resections 15–20 Gy. The radiation is delivered through a cone, usually towards the tumor bed. Some centers routinely irradiate the presacral area where most LR appear [20,21]. A large tumor bed can be difficult to adequately cover with IOERT. Therefore, a few departments give multiple fields, with a risk of overlapping and consequently overdosing problems.

HDR radiotherapy treatment is delivered through parallel catheters in a flexible plastic flap. The size of the flap is adjusted according to the region at risk and is packed onto the curving pelvic surface. The packing will displace the normal organs which thereby avoid radiation damage. It is a disadvantage that the radiation procedure takes nearly an hour, and the penetration depth is only 0.5–1 cm.

Irradiation with iodine 125 seeds is presently not very much applied, and is therefore not further discussed.

In USA and Europe IORT is mostly combined with EBRT, as the dose of a single intraoperative fraction is considered too low [22].

In primary rectal cancer IORT has been given for “locally advanced cases” which may have included T3/T3 +/T4 or N+ cases, i.e. both “intermediary risk” (IR) and high-risk true “locally advanced” (LA) cancers [3] in varying mixtures.

A varying number of patients with locally recurrent rectal cancer (LRRC) have been treated with EBRT for their primary cancer. These patients may benefit from a reduced dose of preoperative reirradiation [22].

Chemotherapy has often been combined with radiotherapy. Various chemotherapy regimens have been used, according to changing standards.

Around 40 reviews on IORT for rectal cancer have been published. All of them claim that IORT has an effect. Three recent reviews have concluded that IORT reduces the frequency of LR [23–25] while a slightly older review claims improvement of both disease-free survival (DFS) and LR rate [26].

Studies on IORT have now been published for nearly 30 years. Still the effect of IORT in rectal cancer treatment is not clear. Here we give a review of the available literature.

Material and methods

Data

Studies of IORT in rectal cancer published after 1989 were collected by searching PubMed and EMBASE and our own database, for studies published in English before 1 June 2013. The keywords IORT, IOERT, EBRT, surgery and locally advanced primary or locally recurrent rectal cancer were used.

Non-IORT studies were preferably selected from randomized studies with chemoradiotherapy including 5FU or capecitabine on locally advanced rectal cancer with the Keywords: EBRT, surgery and locally advanced primary or locally recurrent rectal cancer.

We recorded the preoperative cancer stage and imaging, level of evidence, the preoperative radio- and chemotherapy as well as five-year survival and LR rate.

Papers were excluded when: 1) patients with primary and locally recurrent rectal cancer were analyzed and reported together; 2) IORT was used without EBRT; and 3) actuarial five-year survival or local recurrence rate could not be stipulated. The most recent paper, including the largest number of patients, was picked among several publications from the same department, unless different patient populations were studied. Studies including surgery only were also excluded as preoperative EBRT is considered better than surgery only in LARC, although this is challenged in a recent study from the Mayo Clinic [27].

Unfortunately, most of the available studies on IORT are not suitable for meta-analysis. In a recent review of 29 studies including 3003 patients only five studies on primary cancer and two studies on locally recurrent cancer, including a total of 631 patients, could be used for meta-analysis [25]. These studies were small as only two included more than 50 patients in each arm. The rest of 2372 operated patients were of no value for the evaluation of the effect of IORT.

We have here selected well accepted, large studies with optimal non-IORT treatment which means preoperative irradiation with or without sensitizing 5FU or capecitabine-based chemotherapy.

The available studies are divided into four groups:

1. Randomized studies (level of evidence II) [3];

2. Studies comparing an IORT group with a “control group” (level of evidence III);

3. IORT studies without “control group” (level of evidence V);

4. Contemporary studies without IORT (level of evidence V).

Statistical analysis

As a meta-analysis could not be performed we give in this review the range and mean values for survival and local recurrence rate presented in the different publications. The heterogeneity in design and reporting of the IORT studies implies that weighted means should not be used. We anticipated that a definite effect of IORT would influence the mean values of survival and local recurrence rate compared to those of patients not receiving IORT. Moreover, the results of individual studies with and without IORT should ideally not overlap.

For the comparing studies the IORT and non-IORT results were included in the respective groups.

Results

Primary cancer

Totally, we read 188 abstracts on IORT. Fifteen of 46 relevant papers were selected for this review (flow chart in Figure 1). IORT treatment was performed in 12 centers.

Figure 1. Flow chart showing reason for excluding references from PubMed, EMBASE and own database from inclusion in Tables I and III.

Data from the 15 individual studies are shown in Table I. Totally 1929 patients were given IORT while 2343 were in the non-IORT groups. The comparing studies include considerably fewer patients than the non-comparing ones. Table I also shows that there were slightly more locally advanced cases in the non-IORT/non-comparing group while there were similar numbers of intermediary risk and locally advanced cases in the other three groups. It should be noticed that only a few studies routinely applied MRI for the preoperative diagnosis.

The only randomized study was stopped prematurely. This study is relatively small and therefore considered to be at level of evidence II. Other studies were mostly prospectively recorded consecutive series in which IORT was given on special indications. Some of the non-IORT studies were investigating other factors. These were considered level of evidence V.

The frequency of R0 resections varied from 59% [28] to above 90% [5,29–32]. However, the frequency of R-stages in different studies could not always be compared and was not always reported [11,17,28,33–38]. Thus the results must be interpreted with caution.

Table I. Studies with IORT on primary rectal cancer compared with non-IORT studies. Author with reference number in square brackets. Observation time (Obs) is given in months. Percentages are given for the following parameters: risk stage, surgical R stage, patients given intraoperative radiotherapy (IORT), external radiotherapy (EBRT), and chemotherapy (ChT). The effect of IORT is given as percent of overall survival (OS), cancer specific survival (CSS) or disease-free survival (DFS) and local recurrence (LR) percentage for the total group and R stages.

						Treatment				Survival at 5 years			Local recurrence
Author	No. operated	Risk staging	Imaging	Levels of evidence	R stage %	IORT % Dosage	EBRT % Dosage	CT %	Observation months	OS	CSS	DFS	3 years	5 years	At observation
non-IORT/non-comparing
Larsen [43]	204	LA	CT	V	R0 75 R1 15 R2 10	0	100 46–50 Gy	0	?	total 52 R0 60 R1 31 R2 0				total 15 R0 13 R1 24 R2
Harris [60]	42 M1 6	LA- 100	MRI	V	R0 79	0	74 45 Gy	CRT 74 Adj 38	19	total R0 48 R1-2 33			total 13 R0 R1-2
Park [30]	113	IR -96	CT	V	R0 99	0	100 50 Gy	CRT 75 Adj 58	52	total 85 Ro 85				total 6 R0 6
Lim S-B [29]	581	IR/LA	MRI	V	R0 97 R1 3	0	100 50.4 Gy	CRT 100 Adj 99	58	total 78					total 6
Sauer [5]	415	IR - 68 LA - 6 ? - 21	CT	V	R0 91	0	100	CRT 100	46	total 76 R0 91				total 6 R0
Bosset [35]	506	IR -90 LA 10	CT	V	?+	0	100 45 Gy	CRT 100	63	total 65				total 9
Gerard [36]	375	IR -86 LA -11	?	V	R0-1 93	0	100 45 Gy	CRT 100	81	R0-1 68				R0-1 8
non-IORT comparing
Dubois [17]	69	IR -91 LA- 7	CT	II	R0-1 93	0	100 40–48 Gy	Adj 19	60	total 70 R0 70				total 7
Nakfoor [41]	72	LA -100	?	III	R0 92 R1 8	0	100 50.4 Gy	Adj 25	53		total 77 R0 80 R1-2 0			R0 18 R1-2 83
Pacelli [11]	42	IR -100	CT	III	?	0	0	Adj 52	41	total 57				total 23
Ratto [37]	24-	IR-13 LA-87	MRI	III	?	0	100 45 Gy	CRT 100	74			R0 39 R1 0		R0 43 R1-2 74
Valentini [46]	59-	LA -100	CT/MRI	III	R0 100	0	100 50.4 Gy	CRT 100						R0 19
Ferenschild [14]	93	IR/LA	CT/MRI	III	R0 91	0	100 25–50 Gy	0	25	R0 56 R1-2 0				R0 29 R1-2 100
IORT comparing
Dubois [17]	72	IR-89 LA-7	CT	II	R0-1 100	93 18 Gy	100 40–48 Gy	Adj 25	60	75				total 8
Nakfoor [41]	73	LA-100	CT	III	R0 62	100 10–20 Gy	100 50.4 Gy	CRT 100	53		R0 63 R1 40 R2 14 R1-2 32			R0 11 R1 32 R2 43 R1-2 35
Pacelli [11]	66	IR-100	CT	III	?	100 10 Gy	100 38 Gy	Adj 39	63	total 81				total 7
Ratto [37]	19-	IR-? LA-100 ?	MRI	III	?	100 10–15 Gy	100 45 Gy	CRT 100	74		R0 47			R0 9
Valentini [46]	29-	LA-100	CT/MRI	III	R0 100	100 10–15 Gy	100 50.4 Gy	CRT 100						R0 0
Ferenschild [14]	27	IR/LA	CT/MRI	III	R0 63	100 10G	100 50 Gy		25	R0 66 R1-2 38				R0 28 R1-2 42
IORT non-comparing
Kusters [64]	290	IR-39 LA-61	CT/MRI	V	R0 85 R1 13 R2 2	83 10–17.5 Gy	100 45–50.4 Gy	70	45		total 68 R0 73 R1-2 31			total 13 R0 8 R1-2 38
Kusters [34]	605	IR-71 LA-29	CT/MRI	V	R0 70	100 10–12.5 Gy	100 45–50 Gy	CRT 64 Adj 42	56	total 67 R0 R1-2	R0 74 R1-2 38			total 12 R0 10 R1-2 45
Roeder [31]	243	IR-66 LA-20	CT/MRI	V	R0 92 R1 7 R2 1	100 10–15 Gy	86 16–57 Gy postop 50	86	59					total 8 R0 6 R1-2 18
Palmer [28]	58	LA	CT/MRI	V	R0 59	33	100 50 Gy	CRT 100	64	total 29 R0 34 R1-2 24					R0 47 R1-2 29
Diaz-Gonzales [38]	62	I-83 LA-17	CT	V	?	100 10–15 Gy	100 50.4 Gy	CRT 100 Adj 57	46	total 74				total 6
Huber [39]	38	IR-50 LA-50	?	V	R0 68 R1 16 R2 16	100 15 Gy	100 40.50 Gy T3-poCRT T4-preCRT		26	total 28				total 13
Sadahiro [33]	99	IR-88 LA-12	CT	V		100 15–25 Gy	100 20 Gy	CRT 33	67	total 83				total 2
Mathis [27]	146 Rectum 119 Colon 27	LA-100	?	V	R0 68 R1 19 R2 12	100 7.5–20 Gy	100 50.4 Gy	CRT 100 Adj 40	44	total 52 R0 R2		R0 45 R2 26		total 14 R0 R2
Ratto [37]	43	IR-7 LA-93	MRI	V	R0 88	44 10–15 Gy	100 45 Gy	CRT 100	74			total 39 R0 43 R1 0 R2 0		total 31 R0 26 R1 R2
Valentini [46]	86	LA-100	CT/MRI	V	R0 78	29 10–15 Gy	100 50.4G	CRT 100	31	total 59 R0 68 R1-2 29				total 10 R0 R1-2
Ferenscild [14]	123	IR/LA	CT/MRI	V	R0 85 R1 14 R2 1	22 10 Gy	100 25–52 Gy	0	25	total 50				total 35
Sanfilippo [32]	32	LA-100	CT	V	R0 91	9 15 Gy	100 25–63 Gy	CRT 100 Adj 73	31	total 50 R0				total 20 R0 29

Adj, adjuvant chemotherapy; CRT, chemoradiotherapy; IR, intermediary risk group; LA, locally advanced group; Nadj, neoadjuvant chemotherapy.

Treatment with IORT was heterogenic. Some centers used HDR-IORT [13,14,39] while the majority applied IOERT. Doses were mostly 10–15 Gy depending on amount of residual tumor, while some routinely used 10 Gy [11,14].

Indications for giving IORT varied from treatment to all cases [38] to treatment of only micro- or macroscopic remaining cancer or in cases with a free resection margin less than 1 mm [32], 2 mm [14] or 5mm [40]. Others have not specified the indication for IORT treatment. Some have omitted IORT due to lack of availability of IORT [12]. The frequency of IORT given ranged from 9% [32] to 100% (most studies). Most departments directed IORT towards the tumor bed, some routinely towards the presacral space [31,38].

When IORT was not given to all patients, IORT and non-IORT groups could sometimes be analyzed separately [11,12,14,26,37,41,42].

EBRT also varied to some extent. Most centers gave preoperatively approximately 50 Gy in 1.8–2 Gy fractions, both in the IORT and non-IORT groups, varying from 20 Gy [33] to 57 Gy [31].

Use of sensitizing chemotherapy varied from 0% [14,43] to almost 100% (most studies). Many centers gave adjuvant postoperative chemotherapy varying from 25% [41] to 99% [29].

Survival and LR rate

The mean and limits for survival and LR rate are shown for the IORT and non-IORT studies in Table II. There is no indication of improved survival for patients treated with IORT, neither in the total group, nor in the R0 group, compared to corresponding non-IORT groups. Few R1 and R2 patients are reported. LR rate is slightly higher in the total IORT group, but is slightly lower for patients with R0 resections. For the R1-2 groups a higher LR rate was found in the non-IORT group. It is clear that patients with R0 resections do better for both OS and LR rate compared to those with R1-2 resections.

Table II. Primary cancer. Limits, mean and number of studies giving overall survival and local recurrence for the total group of patients and the separate R stages.

	Limits		Mean		No. studies
Operated	IORT	non-IORT	IORT	non-IORT	IORT	non-IORT
OS total	28–76	52–85	60	72	12	8
R0	34–68	48–91	56	70	3	7
R1	0	31	0	31	1	1
R2	0	0	0	0	1	1
R1-2	0–38	0–33	30	17	3	2
LR total	2–35	5–9	13	8	12	6
R0	0–28	5–43	12	17	8	8
R1		24		24	0	1
R2					0	0
R1-2	18–45	74–100	33	82	3	3

LR, local recurrence at 5 years; OS, overall survival.

The most frequent conclusion of the studies is that achievement of R0 resection is more important than giving IORT [32,34,40,44]. It is also concluded that IORT should be given towards the tumor bed and not routinely towards the presacral area [45]. One of the studies suggested that IORT reduces LR by interaction with subsequent adjuvant chemotherapy [34].

Among the papers with no internal comparison one reported reduction of infield recurrences, although not in R1 resections [32]. Another suggested that IORT reduced infield recurrences by 20–30% [31], whereas one found that a substantial number of presacral recurrences appeared in spite of IORT [45].

The median or mean observation time varied from 25 [14] to 74 months [37], but a number of studies did not give this information.

There is only randomized study [17] among the comparing studies. This study had the highest number of patients included. However, no effect on survival or LR rate was found.

In a small study [14] comparing 11 IORT patients and eight non-IORT patients, improved survival and reduced LR rate were found for patients with R1-2 resections.

A study on 29 patients with R0 resections of locally advanced cancers found low LR rate (0%) after IORT [46]. This is definitely better than the LR rate reported from the same institution on less advanced intermediary risk cancers similarly treated [11].

Locally recurrent cancer

Eighteen papers from 13 centers were selected from originally 48 papers (Figure 1, Table III), including 1174 patients. Also here the comparing studies included fewer patients than the non-comparing ones. In five papers 354 patients with locally recurrent cancer not given IORT, were reported.

Table III. Studies with IORT on locally recurrent rectal cancer compared with non-IORT studies. Author with reference number in square brackets. Observation time (Obs) is given in months. Percentages are given for the following parameters: surgical R stage achieved, patients given intraoperative (IORT), external radiotherapy (EBRT), and chemotherapy (ChT). The effect of IORT is given as overall survival (OS), cancer specific survival (CSS), disease free survival (DFS) and local recurrence (LR) percentage for the total group and R stages.

				Treatment				Survival				Local rerecurrence
Author	No. operated	Level of evidence	R stage %	IORT %	EBRT %	ChT %	Observation months	OS 3 year	OS 5 year	CSS 5 year	DSF 5 year	3 year	5 year	At observation
non-IORT/non-comparing
Lee [62]	45	V	42	0	100 post 45–72 Gy	Adj 88	41		total 49 R0 60 R1-2 43				total R0 22 R1-2 34
Mohiuddin [55]	34	V		0	100	100	24		total 22					total 56
Hansen [57]	185	V	R0 52 R1 48	0	71 50 Gy		15		total 15 R0 55 R1-2 20					R0 18 R1-2
Park [58]	38	V	R0 61 R1-2 39	0	100 50 Gy		31		total 14 R0 35 R1-2 0
Wells [49]	52	V	R0 79 R1 11	0	50 50.4 Gy	39CRT	29		total 17 R0 25 R1 0					total 64 R0 R1
nonIORT / Comparing
Suzuki [47]	64	V	R0 0	0	77 7–74 Gy	?	44		total 7 R1-2 7			R1-2 93
Mannaerts [10]	19	V	R0 37 R1 42 R2 21	0	100 25–50G	0	31	R0 29 R1-2 0	total 7 R0 R1-2			R0 71 R1-2 100	total 86 R0 R1-2
Wiig [12]	48	III	R0 54 R1 21 R2 25	0	100 46–50 Gy	0	?		total 35 R0 60 R1 23 R2 0				total 70 R0 30 R1 75
Hashiguchi [42]	22	III	R0 14 R1 23 R2 64	0	79 50 Gy	?	?		total 0 R0 0 R1-2 0					total 38 R0 R2
IORT/comparing
Suzuki [47]	42	V	R0 0 R1 19 R2 81	100 15–20 Gy	98 7–74 Gy	?	44		total 19 R1-2 19			R1-2 40
Mannaerts [10]	33	V	R0 64 R1 21 R2 20	100 10–17.5 Gy	100 25–50 Gy	0	> 32	R0 63 R1-2 52	total 46 R0 R1-2			R0 11 R1-2 55	total 28 R0 R1-2
Wiig [12]	59	III	R0 31 R1 49 R2 20	100 12–20 Gy	100 46–50 Gy	0	?		total 30 R0 63 R1 28 R2 0				total 50 R0 30 R1 53 R2
Hashiguchi [42]	25	III	R0 24 R1 12 R2 64	70 15–30 Gy	88 50 Gy				total 21					total 11
IORT/non-comparing
Abuchiabe [8]	27	V	30	100 15 Gy	74 46 Gy	0	R0 25 R1-2 65		total 12 R0 29 R1-2 7				total 74 R0 44 R1-2 87
Martinez-Monge [48]	80	V	R1 49 R1-2 51	100 IO/HD/ I125	35 45 Gy	24	48		total 4 R1 6 R2 7				total 74 R1 74 R2 71
Shoup [13]	100	V	R0 64 R1 30 R2 6	100 12–17.5 Gy	37	Adj 58	23			total 39 R0 51 R1-2 14
Lindel [53]	49	V	R0 51 R1 18 R2 31	100 10–20 Gy	94 50.4 Gy	CRT 73	4-150		total 27 R0 40 R1-2 14				total 65 R0 44 R1-2 83
Haddock [22]	607 Rectum 70 Colon 30	V	R0 37 R1 37 R2 26	100 7.5–30 Gy	96 5–65 Gy	CRT 81 Adj 18	44		total 30 R0 46 R1 27 R2 16				total 28 R0 21 R1 R1-2 32
Roeder [31]	97	V	R0 37 R1 33 R2 30	100 10–20 Gy	55 50 Gy	49 Adj 34	51	R1 18 R1-2	total 30 R0 63 R1 R1-2 11			R1 59 R2 82	total 59 R0 32 R1 R2
Kusters [44]	170	V	R0 54 R1-2 46	98 10–17.5 Gy	85 30 Gy 46 50 Gy 54	63	35		total 41 R0 R1-2	R0 58 R1-2 18			total 46 R0 32 R1-2 71
Hashiguchi [42]	37	V	R0 19 R1 17 R2 64	100 15–30 Gy	75 50 Gy	Adj 100			total 13 R0-1 35				R0-1 76
Rahbari [21]	92	V	R0 59 R1 27 R2 14	68 10-20 Gy	30% 50 Gy	CRT 27 Nadj 2	23			total 47 R0 65 R1 10 R2 0				total 25 R0 R1 R2
Palmer [28]	64	V	R0 34 R1 8 R2 52	66	64 50 Gy	CRT 48	63		total 16 R0 40 R1-2 4					R0 14 R1-2 55
Salo [51]	103	V	R0 69 R1 13 R2 18	52 12-15 Gy	100 pre/ post	CRT 26	23		R0 35 R1 23 R2 9
Bedrosian [59]	85	V	R0 76 R1 24	44	27	CRT 27 Adj 53	43		total 36 R0 45 R1 15				total 49 R0 R1
Wiig [52]	150	V	R0 44 R1 39 R2 17	39 12-20 Gy	100 46-50 Gy	11			total 27 R0 52 R1 14 R2 0				total 52 R0 27 R1 63 R2
Kanemitsu [50]	101	V	R0 60 R1 20 R2 20	18 10-20 Gy	41 45-50 Gy	41 CT/CRT				total 32 R0 43 R1 20 R2 10			total 35 R0 30 R1 45 R2 45

CRT, chemoradiotherapy; IR, intermediary risk group; LA, locally advanced, adjuvant chemotherapy; Nadj, neoadjuvant chemotherapy.

The frequency of R0 resections varied from 0% [47,48] to 79% [42,49], being 37% in the largest study on 607 patients from the Mayo Clinic, also including some colon cancers [22]. The frequency of R-stages was not always reported [8,44]. The IORT schedule and indications were heterogenic also in recurrent rectal cancer. The fraction given IORT varied from 18% [50] to 100%. One center used HDR-IORT [51] while the majority applied IOERT. Dosage was nearly always 10–20 Gy, depending on amount of tumor tissue left, while it varied from 7.5 Gy to 30 Gy for previously irradiated patients [22]. Most departments directed IORT towards the tumor bed, a few routinely gave IORT to the presacral area [8,21].

The indications for IORT varied, in some studies based upon the preoperative examinations [52], and in others based upon perioperative findings, with treatment of only micro- or macroscopic remaining cancer [47], or in cases with a free resection margin less than 1 mm [53] or 10 mm [54]. When IORT was not given to all patients, results in IORT and non-IORT groups could sometimes be reported separately [10,12,42,47].

EBRT also varied from one study to another, although most centers gave preoperatively approximately 50 Gy. In recent studies of IORT in recurrent rectal cancer, large fractions (up to 66%) of patients had previously received EBRT in the primary situation [13,22,28,48,51,55]. In such patients the EBRT dose varied from 0 to 74 Gy [13,22,28, 44,47,48,51,55].

Use of sensitizing chemotherapy differed between studies, varying from 0% [8,10,12] to 100% [56] of the patients reported. Some centers gave adjuvant postoperative chemotherapy, varying from 25% [41] to 100% [57]. The chemotherapy regimen used was not reported in some studies [42,47,57–59].

The lack of a generally accepted staging system for recurrent rectal cancer makes it impossible to correctly compare the groups selected for operation. The local recurrences obviously ranged from smaller anastomotic recurrences [8,21], to large infiltrating tumors needing extensive surgery like sacrectomies [10,49,50].

The median or mean observation time varied from 15 to 63 months, but was not given in some studies [12,42,57].

Survival and LR rate

OS and LR rate are shown for the IORT and non-IORT studies in Table IV. IORT does not seem to influence either the OS or LR rate. Unfortunately, few studies specify the results for R1 and R2 resections, especially in the non-IORT group.

Table IV. Locally recurrent cancer. Limits, mean and number of studies giving overall survival and local rerecurrence for the total group of patients and the separate R stages.

	Limits		Mean		No. Studies
Operated	IORT	nonIORT	IORT	nonIORT	IORT	nonIORT
OS total	4–46	0–46	25	19	14	9
R0	29–63	0–60	45	39	10	6
R1	6–28	0–23	19	12	7	2
R2	0–16	0	6	0	7	1
R1-2	4–14	0–43	11	12	3	6
LR total	28–74	70–92	49	81	10	2
R0	21–44	22–30	37	26	8	2
R1	45–88	75	63	75	5	1
R2	45	45	45		1	0
R1-2	32–87	34	65	34	5	1

LR, local rerecurrence at 5 years; OS, overall survival.

The authors frequently stated that it is more important for the outcome to obtain an R0 resection than to give an IORT boost [8,12,15,21,22,44, 51,54,60].

In a study of patients with mostly R2 resections [47] IORT was found to statistically improve both OS and LR rate. One study [42] found improved survival for nine patients with R0-R1 resections with IORT compared to eight patients without IORT. Another study found better OS and LR rate at three years after IORT compared to historical controls [10]. However, they performed more aggressive surgery and had twice as many R0 resections in the IORT group. The only comparing study which did not report any effect of IORT [12] included few patients in each R-stage.

Discussion

In this overview of IORT for locally advanced and locally recurrent rectal cancer, we have included reports from altogether 36 studies of various type and quality. As we do not have results from large randomized trials it is difficult to draw definite conclusions regarding the question: Does IORT work at all? However, by giving the summarized results as mean with variations, comparisons between groups may give important indications of effects on outcome.

By calculating the mean and variation of results from available studies we have found clear differences in survival and local failure between primary cancers and locally recurrent cases, as well as for R0 resections versus R1-2 resections. This is according to previous reports, and shows that this approach may be useful to investigate differences in outcome between groups.

Our main conclusion is that no definite improvement by IORT has been documented. In primary cancer, there was neither any effect on OS nor LR rate for patients with R0 resections or the total group (including R0, R1 and R2 resections). For R1 and R2 patients no conclusion regarding OS could be made. The mixed R1-2 results possibly suggested an effect of IORT on LR rate. For locally recurrent cancer there was no effect of IORT on OS, neither for the total group, nor for the separate R-stages. This also applies to the LR rates.

The results clearly show the importance of separately analyzing and reporting primary and locally recurrent cancers and their specific R-stages.

Are the IORT and non-IORT studies comparable? Using the TNM staging system studies in primary cancer should be relatively comparable with regard to extent of cancer. However, magnetic resonance imaging (MRI) or computed tomography (CT) was in many of the studies not routinely used and the preoperative diagnosis therefore uncertain [3]. In locally recurrent cases, no generally accepted staging system exists. Therefore, comparisons of groups of patients are even more uncertain than in primary cases. Differences in selection of patients and quality of surgery can profoundly influence the results in such cases. For some of these studies good results might be partly explained by a high frequency of anastomotic recurrences [8,21].

The reports on primary cancer include patients with intermediary risk and locally advanced cancers. However, preoperative examinations suggested less locally advanced cancers in the IORT/non-comparing group. In contrast, the IORT comparing group included patients with a higher R-stage. This was supposed to be the reason why there was hardly any LR in the non-IORT R0 group of two studies. If there was any effect of IORT this could only be small as the presence of LR in R1 resected cases shows that IORT cannot eradicate even microscopically remaining cancer [21,43]. Treatment with IORT might have reduced the inter-study variation in results. For IORT studies this was similar to that of the non-IORT studies. A small effect could be hidden by the heterogeneity in technique (HDR/IOERT), dosage, localization [20,31,34], definition of “close” margin [14,32,40], differences in frequency of IORT given [13,28,41] and inclusion of varying numbers of previously irradiated patients in the studies of locally recurrent cases.

The inclusion of M1 stages in some reports [21,42,49,60,61] confounds the evaluation of IORT effect even more.

In primary cancers patients with R0 resections will have a low rate of local recurrences and IORT effect can therefore be difficult to identify. Our results are in line with this and strongly support the previous suggestion that IORT should not routinely be given to R0 resections of cancers with intermediary risk [17,25]. In R1 resections with microscopic remaining cancer the LR frequency will be higher and this should therefore be the ideal group for the identification of a potential IORT effect. Also patients with R0 resections of locally recurrent cancer with a relatively high rate of local rerecurrences, could be expected to improve with IORT treatment. Similarly, one would expect that cases with R1 resections of local recurrences could be susceptible to IORT. It is therefore unfortunate that the R-stage is not specified in several of the studies of both the IORT and non-IORT groups of primary cancer as well as local recurrent cancer [11,35–37,43,56,59,62,63]. As locally recurrent rectal cancer grow in less well-oxygenated scar tissue, and theoretically can be radioresistant in cases with previous irradiation, one might imagine that all cases with locally recurrent cases would need higher doses of IORT. This was not discussed in any study. The fact that the Mayo Clinic in their large studies consistently reports better results than most other centers for R2 resected LRRC could be due to their use of 20 Gy IORT [22], which is higher than used by most others.

The R-staging can be difficult [64] and therefore inaccurate, and is particularly problematic in locally recurrent rectal cancer. This can explain part of the variation of results within separate R-stages and also the higher frequency of LR in R0 stage of LRRC compared to primary cancer. In fact, none of the studies declare the number of microscopic slides examined for the R-staging.

The most frequent conclusion reported is that IORT is not able to improve the results after a R1 resection to the level obtained after a R0 resection. This applies for both primary [32,40,44,46] and recurrent cancer [8,12,15,22,51,54,60]. However, conclusions in various studies are contradictory [8,13].

A reappraisal of the conclusions of comparing studies showing statistical significant improvement with IORT boost suggests that these results must be interpreted with caution.

Among the studies on primary cancers the only randomized study [17] neither found effect on survival nor on LR rate. This might be due to a low number of LR. This conclusion is in contrast to the conclusion made by other authors, based on consecutive, similar patient groups [31,33,34,38].

All other studies included low number of patients where differences in tumor biology or inaccurate staging might profoundly influence the results. One study included less than 10 patients in each comparing group [42], two others less than 20 [10,14] and a fourth less than 30 patients in one of the groups compared [26]. Also one of the studies had surprisingly low estimated three- (11%) and five-year OS (0%) in the historic control non-IORT group [10] and twice as many R0 resection in the aggressively operated IORT group. In fact this study is often cited as proving the effect of IORT [65]. However, one of the studies reported no LR after IORT [26] which is better than for patients with less extensive tumors reported from the same department [11]. The results also seem inconsistent as one study on locally advanced cancers found improvement in LR rate after R0 resection [26], while another in contrast found improved OS and LR rate after R1-2 resections [14]. For locally recurrent cases one study found effect on OS of the total group [10], another on OS of R0-1 resections [42] and a third found improvement of both OS and LR after R2 resections [47].

What is the biological effectiveness of radiation therapy given as IORT? Interestingly, the recent paper from the Mayo Clinic [22] suggested that the additive effect of EBRT and IORT may be lower than previously estimated [6]. It was suggested that the combined dose of EBRT 50 Gy plus intraoperative irradiation 12 Gy given four weeks later, might be equivalent to 66 Gy in 2 Gy fractions, and that EBRT reirradiation of 30 Gy plus 12 Gy IORT immediately thereafter for previously irradiated patients is equivalent to 63 Gy (Haddock pers. comm.).

Conclusion

The studies here referred, including both IORT and non-IORT-treated patients, suffer from a lack of standardization in the presentation of the results, making the groups difficult to compare. The results reported for the IORT and non-IORT groups, both for primary and recurrent cancer and their R-stages, are difficult to interpret. The wide variation in results for IORT-treated patients is within the same limits as for the non-IORT-treated patients. Also the partly contradicting conclusions of the IORT papers do not provide any convincing evidence of an IORT effect.

Interestingly, a new model for estimation of the combined effect of EBRT and IORT has suggested less additive effect than previously assumed and therefore theoretically a possibly smaller benefit from IORT may be expected.

After 30 years of treatment with IORT it is clear that IORT unfortunately has not had a great impact on treatment results. It still cannot be completely excluded though, that there might be an effect on LR rate.

Only large randomized studies can establish whether IORT is of any therapeutical value. As far as we know, no such study is ongoing or planned. Is IORT then of clinical relevance?

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.