Introduction: The UK Medical Research Council CLASICC trial assessed the safety and efficacy of
laparoscopically assisted surgery in comparison with open surgery for colorectal cancer. The results of
the 5-year follow-up analysis are presented.
Methods: Five-year outcomes were analysed and included overall and disease-free survival, and local,
distant and wound/port-site recurrences. Two exploratory analyses were performed to evaluate the effect
of age (70 years or less, or more than 70 years) on overall survival between the two groups, and the effect
of the learning curve.
Results: No differences were found between laparoscopically assisted and open surgery in terms of
overall survival, disease-free survival, and local and distant recurrence. Wound/port-site recurrence rates
in the laparoscopic arm remained stable at 24 per cent. Conversion to open operation was associated
with significantly worse overall but not disease-free survival, which was most marked in the early followup period. The effect of surgery did not differ between the age groups, and surgical experience did not
impact on the 5-year results.
Conclusion: The 5-year analyses confirm the oncological safety of laparoscopic surgery for both
colonic and rectal cancer. The use of laparoscopic surgery to maximize short-term outcomes
does not compromise the long-term oncological results. Registration number: ISRCTN74883561
(http://www.controlled-trials.com).
Presented to the National Cancer Research Institute Cancer Conference, Birmingham, UK, October 2007
Paper accepted 27 April 2010
Published online 13 July 2010 in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.7160
Introduction
1639
site, presence of liver metastases, preoperative radiotherapy and surgeon) and other prognostic factors (age, sex
and Dukes stage). Sensitivity analyses were performed to
assess the impact of exclusions for DFS on OS. Sensitivity
analyses were also performed to assess the effect of conversions (from laparoscopic to open surgery) on the results.
As sensitivity analyses or adjustment for sex, age, Dukes
stage and stratication factors made little difference to
conclusions, differences between treatment estimates are
presented from unadjusted analyses.
Two additional exploratory analyses were performed.
The rst investigated the effect of surgery on OS for
patients who were aged 70 years or less at the time of randomization and for those who were over 70 years old. The
second analysed only those patients recruited by surgeons
who randomized more than 20 patients. In this analysis,
patients were categorized into two groups, the rst and
last half of randomized cases per surgeon, and the effect of
surgery on OS was examined in these separate groups to
investigate whether the experience gained during the trial
had an impact on the results.
All hypothesis tests were at the 1 per cent signicance
level (two-sided) for the 5-year endpoints and were performed using the intention-to-treat and actual treatment
populations (analysed according to treatment actually
received: open, laparoscopic or laparoscopic converted
to open surgery). All statistical analyses were performed
using SAS version 9.1 (SAS Institute, Cary, North Carolina, USA).
Results
Overall survival
The 5-year OS rate for all patients was 579 (95 per cent
condence interval (c.i.) 544 to 615) per cent, with 109
deaths in the open arm and 213 in the laparoscopic arm.
Overall cause of death was similar in the two arms. There
was no difference in the 5-year OS rate between the
two groups (581 per cent for open versus 579 per cent for
laparoscopic surgery; difference 02 (95 per cent c.i. 76
to 73) per cent; log rank statistic = 0037, P = 0848).
www.bjs.co.uk
1640
Follow-up
Allocation
Analysis
Enrolment
CONSORT diagram depicting allocation of patients in 5-year follow-up analysis. OS, overall survival; DFS, disease-free survival.
*Patients not known to have died and whose last follow-up information was less than 5 years from randomization. Also time to local,
distant and wound/port-site recurrence. More than one reason may be given per patient
Fig. 1
www.bjs.co.uk
Proportion surviving
10
09
08
07
06
05
04
03
02
01
0
1641
Open
Laparoscopic
Conversion
6
12
18
24
30
36
42
48
54
60
65
98
22
Fig. 2
Disease-free survival
A total of 641 patients (212 randomized to open and 429
to laparoscopic surgery, 315 with colonic and 326 with
rectal cancer) were eligible to be included in the DFS
analyses, and time to local, wound/port-site and distant
recurrence analyses.
The 5-year DFS rate for all patients was 564 (95 per cent
c.i. 524 to 603) per cent, with no difference between
the two surgical techniques: 586 per cent for open versus 553 per cent for laparoscopic surgery (difference 34
(95 per cent c.i. 118 to 50) per cent; log rank statistic =
0492, P = 0483). DFS for patients with colonic and rectal
cancer was 600 (95 per cent c.i. 544 to 656) per cent and
529 (473 to 585) per cent respectively.
There was no difference in DFS for patients with either
colonic or rectal cancer with respect to the randomized
procedure. For colonic cancer, the 5-year DFS rate was
640 per cent for open versus 576 per cent for laparoscopic
surgery (difference 64 (95 per cent c.i. 179 to 52)
per cent; log rank statistic = 0712, P = 0399). For rectal
cancer, the 5-year DFS rate was 521 per cent for open versus 532 per cent for laparoscopic surgery (difference 11
(95 per cent c.i. 112 to 134) per cent; log rank statistic
= 0004, P = 0953).
Overall, there was no difference between the two techniques for any stage of colonic or rectal cancer. The
non-signicant trend for improved DFS after laparoscopic
2010 British Journal of Surgery Society Ltd
Published by John Wiley & Sons Ltd
Local recurrences
The overall local recurrence rate at 5 years was 101
(95 per cent c.i. 75 to 127) per cent, with no difference between the two procedures (87 per cent for open
versus 108 per cent for laparoscopic surgery; difference
21 (95 per cent c.i. 73 to 32) per cent; log rank
statistic = 0285, P = 0594). In patients with rectal cancer, local recurrence rates following anterior resection and
abdominoperineal resection were 89 (95 per cent c.i. 49
to 128) per cent and 177 (84 to 269) per cent respectively, giving a difference in the 5-year local recurrence rate
of 88 (95 per cent c.i. 13 to 188) per cent. In patients
undergoing anterior resection, the previously reported differences in CRM positivity rates did not translate into a
difference in the 5-year local recurrence rate: 76 per cent
for open versus 94 per cent for laparoscopic surgery (difference 18 (95 per cent c.i. 99 to 63) per cent; log
rank statistic = 0110, P = 0740). Similarly, there was no
difference in local recurrence rates between the two techniques in patients undergoing abdominoperineal resection
(data not shown).
Distant recurrences
Some 111 distant recurrences were recorded within 5 years
of randomization, giving an overall distant recurrence
rate of 209 (95 per cent c.i. 174 to 244) per cent. No
difference was observed between the two surgical techniques: distant recurrence rate of 206 per cent for open
versus 210 per cent for laparoscopic surgery (difference
04 (95 per cent c.i. 78 to 71) per cent; log rank
statistic = 0052, P = 0820). There was no difference
www.bjs.co.uk
1642
in distant recurrence rates in patients with rectal cancer undergoing either anterior resection (219 per cent for
open versus 219 per cent for laparoscopic surgery; difference 002 (95 per cent c.i. 128 to 129) per cent; log
rank statistic = 0027, P = 0869) or abdominoperineal
resection (408 per cent for open versus 357 per cent for
laparoscopic surgery; difference 51 (210 to 313) per
cent; log rank statistic = 0092, P = 0762).
The 5-year distant recurrence rate for patients converted
from laparoscopic to open operation was not signicantly
worse even after adjustment for the stratication factors of
sex, age and Dukes stage (P = 0679).
Wound/port-site recurrences
No further port-site recurrences were reported between
3 and 5 years after randomization. Overall, there was
one wound/port-site recurrence in the open arm and
nine in the laparoscopic arm (laparoscopic wound/portsite recurrence rate 24 per cent), of which only one
was highlighted as being a true port-site rather than an
extraction-site recurrence.
Exploratory analyses
In the exploratory analysis for age, 410 patients were
aged 70 years or less at the time of randomization, with
134 assigned to open and 276 to laparoscopic surgery.
There was no difference in the 5-year OS rate for the two
techniques: 650 per cent for open versus 677 per cent for
laparoscopic surgery (difference 27 (95 per cent c.i. 73
to 126) per cent; log rank statistic = 0364, P = 0547).
Similarly, there was no difference in the 384 patients older
than 70 years: 510 per cent of 134 patients having open
versus 470 per cent of 250 patients undergoing laparoscopic surgery (difference 40 (149 to 69) per cent; log
rank statistic = 0171, P = 0679).
In the exploratory analysis that looked at experience
gained within the trial, and which included only those
patients recruited by surgeons who randomized more than
20 patients, 326 patients constituted the rst half and
331 the second half of patients recruited. There was no
difference in 5-year OS between the open and laparoscopic
arms for either group of patients.
Discussion
1643
www.bjs.co.uk
Acknowledgements
1644
Collaborators
www.bjs.co.uk
1645
www.bjs.co.uk