Prevention by Aspirin of Colorectal Adenoma Recurrence: Some Advances and Latest Results of the APACC Trial
- First Online:
- Cite this article as:
- Benamouzig, R., Uzzan, B., Deyra, J. et al. Curr Colorectal Cancer Rep (2011) 7: 33. doi:10.1007/s11888-010-0079-5
- 42 Views
In a randomized controlled trial (RCT), 272 patients were assigned to lysine acetylsalicylate 160 mg/day (n = 73) or 300 mg/day (n = 67) or placebo (n = 132). The primary end points were adenoma recurrence and adenomatous polyp burden (APB) at year 1 or 4 (last colonoscopy) and at year 4. At last colonoscopy, APB tended to be lower under aspirin 160 mg or at either dose compared with placebo (P = 0.06 and 0.07, respectively). At year 4, 55 patients had received aspirin 160 mg/d, 47,300 mg/d, and 83 placebo. APB and proportions of patients with at least one recurrent adenoma were similar in both groups. A personal history of adenomas and an initial APB higher than 10 mm predicted recurrence. Among 219 adenomas from 136 patients, 128 adenomas (58%) from 59 patients strongly expressed COX-2 assessed by immunohistochemistry, mainly adenomas larger than 10 mm (84/129 vs 44/90; P = 0.02) and adenomas showing high-grade dysplasia (22/29 vs 104/188; P = 0.04). Deep stromal initial expression of COX-2 predicted recurrence (P = 0.04). Protection by aspirin was mainly observed in patients in whom COX-2 initial expression was low (RR: 0.59; 95% CI = 0.39–0.90; P = 0.02). Aspirin decreased adenoma recurrence significantly at 1 year, marginally at year 1 or 4, and not at year 4, possibly due to attrition but also to a differential effect according to polyp natural history. In a recent patient-level meta-analysis including the four published RCTs, aspirin significantly decreased adenoma risk by 17% and risk of advanced lesions by 28%.
KeywordsColorectal adenomaChemopreventionAcetylsalicylic acidRandomized controlled trialPlaceboCOX-2 expression
Cohort and case–control studies show that long-term use of low-dose aspirin reduces the risk for colorectal cancer (CRC) by approximately 40% to 50% [1–8, 9•, 10••, 11, 12, 13••, 14], decreasing disease-specific and overall mortality [15•]. Aspirin also reduces the risk of colon adenomas, precursors of most CRCs. In a systematic review, aspirin reduced the incidence of colon adenomas and CRC, especially when used for more than 10 years, with a dose-related increase in gastrointestinal complications . In vitro and animal studies also found a protective effect of aspirin on experimental colon carcinogenesis [1, 17]. Four randomized placebo-controlled clinical trials of aspirin for prevention of CRCs or adenomas have been reported. The oldest trial compared 325 mg of aspirin every other day with placebo among healthy male physicians and found no effect of aspirin [18, 19]. The next trial compared 325 mg of aspirin per day with placebo among 635 patients with a history of CRC; aspirin significantly reduced adenoma recurrence by 35% [20••]. A third trial compared 81 and 325 mg of aspirin per day to placebo in 1,121 patients with a history of one or more colorectal adenomas, and found a reduction in recurrences, statistically significant only for the lower dose of aspirin (relative risk = 0.81; 95% CI = 0.69–0.96; P = 0.04) . The fourth trial was performed in 945 patients with adenomas larger than 5 mm. After 3 years of follow-up, 99 of 434 patients (23%) receiving aspirin 300 mg/day recurred, compared with 121 of 419 patients (29%) receiving placebo (relative risk: 0.79; 95% CI = 0.63–0.99; P = 0.04) [22••]. The efficacy of aspirin was confirmed by a meta-analysis of all relevant observational studies of prevention of CRC and of two randomized controlled trials (RCTs) .
The preventive effect of aspirin on adenoma recurrence is now established, but its use in clinical practice is far from widespread because information is still lacking on which subgroups of patients would benefit most from aspirin according to their benefit/risk ratio, the optimum dose, and the duration of treatment.
COX-2 and Aspirin Effect
Cyclooxygenase-2 (COX-2) plays an important part in early colon carcinogenesis. COX-2 expression was detected in intestinal adenomas in the azoxymethane rodent model and in the Min mouse. Disruption of the COX-2 gene or selective COX-2 inhibition dramatically reduced polyp growth in APC delta716 knockout mice as well as Min mouse. COX-2 is expressed at low levels in normal colon but at high levels in 40% to 100% of CRCs and more than 80% of CRC metastases. COX-2 expression was related to unfavorable pathologic findings and TNM stage in several series, and was an independent prognostic factor in some large series but not all. The percentage of adenomas expressing COX-2 varied among studies, from 0% up to 90%. COX-2 expression was associated with size and degree of dysplasia of adenomas in some series. Nonsteroidal anti-inflammatory drugs (sulindac, coxibs) are associated with a reduced risk for adenomas in patients with familial adenomatous polyposis and sporadic adenoma recurrence [24•, 25•, 26•, 27•, 28•]. The chemopreventive effect of aspirin might be related to inhibition of COX-2. Recently, in two large prospective cohorts, regular use of aspirin was found to reduce the risk of CRCs overexpressing COX-2 but not of cancers with weak or absent COX-2 expression [29••].
The APACC Approach
Only one final colonoscopy was performed in the published studies. Thus, the preventive effect of aspirin on missed adenomas could not be distinguished from its true preventive effect. These considerations were even more valid in 1996. Thus, a double-blind placebo-controlled RCT of aspirin (160 or 300 mg lysine acetylsalicylate per day) on secondary prevention of colorectal adenomas after colonoscopic clearance of pre-existing polyps was planned in a population at high risk of recurrence. The 1-year results have been published previously . The final 4-year results on adenoma recurrence and adenomatous polyp burden (APB) are now submitted for publication and presented here. The pathological arm of the APACC study also aimed to assess the overall impact of COX-2 expression on adenoma recurrence independently of aspirin use, and to examine whether or not the preventive effect of aspirin predominated in patients with adenomas highly expressing COX-2.
APACC Methods and Patients
Randomization was balanced every four patients: one patient received 160 mg of aspirin, one 300 mg, and two placebo. Lysine acetylsalicylate was administered as one sachet to be diluted in water. Patients, staff in the APACC coordination center, and study investigators were unaware of treatment assignment. The coordination center was responsible for randomization, patient allocation, preparation, and distribution of treatment packages. Information about compliance and side effects was collected every 4 months by a senior gastroenterologist. All patients gave written informed consent. The trial was completed at the end of September 2005. It was declared to ClinicalTrials.gov and registered as NCT00224679.
The statistical analysis was planned to assess first the effect of either dose of aspirin (160 or 300 mg/d) and placebo and, only in case of a statistically significant overall effect of aspirin, to compare each dose separately with placebo. Patients who experienced the 4-year colonoscopy either in due time (from 6 months before to 6 months after the 4-year visit) or not were also analyzed. Assuming an annual recurrence rate of adenomas of about 10% with a 4-year recurrence rate estimated to 43%, and a dropout rate of 15% over the 4-year period, this trial was designed to have 80% power to detect a reduction of 40% in recurrence rate (two-sided test, 5% significance level). With these assumptions, the recruitment target was 267 patients.
Biopsies were fixed in formalin, then embedded in paraffin. For each sample, one section was stained with hematoxylin/eosin and examined by two independent “blinded” pathologists. Each adenoma was classified as tubular (n = 138; 63%), villous or tubulo-villous (n = 81; 37%) (WHO). The degree of cytological dysplasia was classified as mild, moderate (low grade, n = 188; 86%), or severe with or without carcinoma in situ (high grade, n = 29; 13%). Disagreements between the two pathologists were solved by discussion. Expression of COX-2 was determined by immunohistochemistry using VECTASAIN Elite ABC kit (Vector, Burlingame, CA) and goat polyclonal-specific antibody directed against COX-2 (Santa Cruz Biotechnology, CA). Formalin-fixed paraffin-embedded sections of human colon carcinomas showing strong COX-2 immuno-staining were used as positive controls, whereas omission of the primary antibody served as negative control. COX-2 staining was graded by two independent observers blinded to outcomes. For each adenoma, COX-2 expression was scored separately in epithelial cells, superficial (below luminal surface), and deep stromal cells (within the body of adenoma), then a global score was built up. Epithelial cells were scored from 0 (no staining) to 3 (intense staining in more than 50% of cells). Superficial and deep interstitial cells scored 0 (no staining), 1 (staining of one focus of less than 20 cells), 2 (staining of more than one focus of less than 20 cells), 3 (staining of one focus of more than 20 cells), and 4 (staining of more than one focus of more than 20 cells). COX-2 expression was considered to be low if the score was 0 or 1, and to be high if it was 2 or more. The overall score corresponded to the highest of the three partial scores.
Among the 291 patients who entered the 4-week run-in period, 272 took at least 80% of 300-mg daily aspirin dispensed and were randomized. Seventy-three patients were allocated to aspirin 160 mg daily, 67 to aspirin 300 mg daily, and 132 to placebo (Fig. 1). Baseline characteristics of patients were similar among the aspirin and placebo groups at entry, except for a higher proportion of adenomas larger than 10 mm under aspirin. At baseline colonoscopy, only 13 out of the 668 polyps excised from the 272 patients were lost (2%). Among the 655 excised polyps that benefited from pathological examination, 549 were adenomatous and 85 hyperplastic.
APACC Main Results
Among the 272 randomized patients, 238 (87.5%) had a colonoscopy 1 year after enrolment. At 4 years, 185 patients (68% of the initial population) had a third colonoscopy, in due time or not: 55 received aspirin 160 mg, 47 aspirin 300 mg, and 83 placebo (Fig. 1). Six patients who had no year 1 colonoscopy could be included in the final analysis since they had year 4 colonoscopy. Only 150 patients (55% of the initial population) had a 4-year colonoscopy in due time and did not present any major protocol violation: 42 received aspirin 160 mg, 40 aspirin 300 mg, and 68 placebo. Among the 122 patients who had at least one major protocol violation, 32 received aspirin 160 mg, 27 aspirin 300 mg, and 63 placebo. Among the 87 patients who did not complete year 4 colonoscopy, 29 withdrew from the trial for adverse event, 25 for noncompliance, 13 for other reasons, and 20 were lost to follow-up. More patients dropped out in the placebo group than in the aspirin group (49/132 or 37% vs 38/140 or 27%; P = 0.08). This might be due to a higher noncompliance in the placebo group (18/49 or 37% vs 7/38 or 18%; NS).
At last follow-up colonoscopy (1-year or 4-year) (Table 1), 244 patients could be analyzed: 32 out of 68 (47%) who had received aspirin 160 mg had at least one recurrent adenoma, compared with 33 out of 60 patients (55%) who had received aspirin 300 mg and to 62 out of 116 under placebo (54%) (NS). APB was 6.3 mm among the 116 patients under placebo, 3.6 mm among the 68 patients under aspirin 160 mg, 4.7 mm among the 60 patients under aspirin 300 mg, and 4.1 mm among the 128 patients who received either dose of aspirin (P values of 0.06 for the comparison between aspirin 160 mg and placebo and 0.07 for the comparison between either dose of aspirin and placebo). Mean number of recurrent adenomas per patient was 1.5 under placebo, 0.9 under aspirin 160 mg, 1.2 under aspirin 300 mg, and 1 under either dose of aspirin (P values of 0.05 for the comparison between aspirin 160 mg and placebo and 0.07 for the comparison between either dose of aspirin and placebo). Aspirin at either dose did not decrease the proportions of patients with at least one adenoma larger than 5 or 10 mm. Proportions of patients with at least one recurrent advanced adenoma were similar under aspirin 160 mg (18%), aspirin 300 mg (10%), and placebo (16%).
Among the 185 patients who had a third colonoscopy at 4 years, 15 out of 55 who took aspirin 160 mg (27%) had at least one recurrent adenoma, compared with 27 out of 47 (58%) who took aspirin 300 mg and 33 out of 83 (40%) under placebo (NS) (Table 1). APB was similar in the three groups: 2.1 mm for aspirin 160 mg, 4.3 mm for aspirin 300 mg, and 3.4 mm for placebo (NS). Mean number of adenomas was significantly lower in patients under aspirin 160 mg than in those under placebo: 0.4 versus 0.8 (P = 0.025).
By multivariate analysis, final recurrence was predicted by male sex (RR: 0.67; 95% CI: 0.42–1.04; P = 0.06), old age (RR: 1.44; 95% CI: 1.00–2.07; P = 0.07), a personal history of adenoma before inclusion (RR: 1.58; 95% CI: 1.13–2.22; P = 0.01), an initial APB higher than 10 mm (RR: 1.45; 95% CI: 0.97–2.17; P = 0.06), the presence of at least three adenomas initially (RR: 2.22; 95% CI: 1.64–3; P = 0.0001), and 1-year recurrence (RR: 1.76; 1.25–2.46; P = 0.001).
Very few serious adverse events occurred during the trial: one death and one myocardial infarction under placebo, and one stroke under aspirin. Twenty-four patients had a minor bleeding episode under aspirin (20 under placebo; no major bleeding in both groups).
Baseline colonoscopy diagnosed a total of 219 adenomas from the 136 patients whose data on COX-2 expression could be analyzed (half of the patients of the RCT). The main characteristics of the COX-2 population were similar to those of the RCT, except for a higher prevalence of tubulo-villous adenomas (70/136 or 51.5% among the COX-2 population vs 112/272 or 41% among the RCT population; P = 0.05). The COX-2 population could be considered as representative of the RCT population. Overall COX-2 expression was high in 128 adenomas (58%). Low overall COX-2 expression was found in 77 patients (57%) and high expression in 59 (43%). COX-2 was strongly expressed in epithelium in 77 adenomas and in deep stroma in 54 adenomas. COX-2 staining predominated on epithelial cells in 142 adenomas and on deep interstitial cells in 77 adenomas. Staining was heterogeneous from one crypt to another and inside the same crypt. No immuno-reactive COX-2 was detected in normal mucosa from controls or adjacent to polyps.
More polyps with high-grade dysplasia strongly expressed overall COX-2, compared to polyps with low-grade dysplasia (22/29 or 76% vs 104/188 or 55%; P = 0.04). A strong overall expression of COX-2 was present in similar proportions among adenomas from proximal or distal colon (52% vs 60%). Significantly more adenomas with diameters superior to 5 or 10 mm strongly expressed COX-2 (99/154 or 64% vs 29/65 or 45% and 84/129 or 65% vs 44/90 or 50%; P = 0.007 and 0.02, respectively). Numbers of adenomas were significantly higher among patients with at least one adenoma strongly expressing COX-2 in deep stroma than among those with low COX-2 expression (118 from 46 patients vs 169 from 90 patients). APB was significantly higher among patients highly expressing COX-2 either in epithelium or in deep stroma. Univariate analyses showed direct associations between COX-2 epithelial expression and APB (RR: 6.6; P = 0.001), polyp size (RR: 3.1; P = 0.001), and between COX-2 deep stromal expression and APB (RR: 4.8; P = 0.001). By multivariate analysis, epithelial COX-2 expression correlated with adenoma size (P = 0.001).
Of the 136 patients included in the COX-2 analysis, 43 had received 160 mg of aspirin daily, 36, 300 mg/d, and 57 placebo. Results of both doses of aspirin were pooled. Seventy-two patients had recurrent adenomas and 64 had no recurrent adenoma during the 4-year study. Proportions of patients with a strong initial overall expression of COX-2 were similar between patients who had or had not recurrent adenomas at either colonoscopy (30/72 vs 29/64). Proportions of patients strongly expressing epithelial COX-2 initially were also similar among those with and without recurrence of adenomas at either colonoscopy (33/72 vs 29/63). By contrast, deep stromal initial expression of COX-2 predicted recurrence of adenomas: 30/72 patients (42%) with recurrent adenomas strongly expressed deep stromal COX-2 initially compared with 16/64 (25%) without recurrent adenoma (P = 0.04). In patients assigned to aspirin at either doses, deep stromal initial expression of COX-2 also predicted recurrence: 16/38 patients (42%) with recurrent adenoma strongly expressed COX-2 initially compared with 7/41 (17%) without recurrent adenoma (P = 0.01).
Aspirin was associated with lower adenoma recurrence compared with placebo among patients with low overall initial COX-2 expression (18/43 or 42% vs 24/34 or 71%; P = 0.02) or low epithelial COX-2 expression (17/42 or 40.5% vs 22/31 or 71%; P = 0.01). A similar trend was observed for low stromal COX-2 expression (22/56 or 39% vs 20/34 or 59%; P = 0.07). By contrast, among patients who highly expressed overall (n = 59), epithelial (n = 62), or deep stromal COX-2 (n = 46) at the initial colonoscopy, the proportions of recurrences were similar between the aspirin and placebo groups.
A multivariate analysis performed on the final year 1 or 4 colonoscopy found that low BMI (OR: 2.4; 95% CI: 1.08–5.3; P = 0.03), personal history of adenoma (OR for no history: 0.23; 95% CI: 0.08–0.66; P = 0.006), tubulo-villous adenoma (OR for absence: 0.42; 95% CI: 0.19–0.94; P = 0.03), and deep stromal COX-2 expression (OR: 0.36; 95% CI = 0.16–0.85; P = 0.02) predicted adenoma recurrence.
Whether COX-2 expression is increased in the epithelial or stromal component of adenomas is still debated. The balanced expression in both types of cells was previously described in series of sporadic adenomas, in familial adenomatous polyposis, and in serrated adenomas. However, in most studies, COX-2 expression was reported predominantly or exclusively in the epithelium. The nature of stromal cells expressing COX-2 remains uncertain. Preliminary data suggested a macrophage expression but more recent data showed a myofibroblastic origin. In this study, deep stromal but not epithelial initial expression of COX-2 predicted adenoma recurrence. This suggests a role of these stromal cells in early colon carcinogenesis and would indicate that carcinogenesis is closely associated with the transformation of normal stroma into a “reactive” stromal phenotype.
Aspirin would be expected to preferentially reduce the recurrence of tumors overexpressing COX-2. The preventive effect of low-dose aspirin on the incidence of recurrent adenomas did not predominate in patients whose initial expression of COX-2 was high but in those whose initial COX-2 expression was low [32••]. In a combined analysis of two studies on 636 incident CRCs, 67% of patients had moderate or strong COX-2 expression. Aspirin significantly reduced the proportion of CRCs overexpressing COX-2 (RR = 0.64; 95% CI: 0.52–0.78), whereas it had no effect on tumors with weak or absent expression of COX-2 . This discrepancy might be explained by differences in timing of assessment of COX-2 expression, performed in carcinomas after aspirin treatment and in adenomas before treatment. Results on COX-2 should be interpreted cautiously (small numbers of patients at final colonoscopy). However, because COX-2 overexpression correlates with more advanced histology, adenoma size, and adenoma recurrence, it may be a marker of a significant “field” defect predisposing to recurrence and corresponding to a later stage of carcinogenesis less susceptible to aspirin. Patients with COX-2 high baseline adenomas might be farther down the pathway toward carcinogenesis and less likely to respond to aspirin.
Aspirin Effect: Time and Dose Variation
In the APACC trial, the effect of aspirin on adenoma recurrence or APB at last follow-up colonoscopy (year 1 or 4) was of borderline statistical significance. The final 4-year results involving 185 patients out of 272 randomized did not show any effect of aspirin. In addition, final 4-year adenoma recurrence was significantly predicted by a personal history of adenoma before inclusion, by the presence of at least three adenomas at inclusion and, most importantly, by 1-year recurrence.
The discrepancy between results at 1 (significant prevention by aspirin) and 4 years could result from the lack of statistical power of final analysis (numerous dropouts; 32%). The main reasons for this were the long duration of follow-up and the unsatisfactory need for patients to restrain from using aspirin outside the trial. In the 1-year analysis of the 185 patients included in the 4-year analysis, aspirin at either dose significantly reduced the proportion of patients with at least one recurrent adenoma (P = 0.05) and APB (P = 0.005). This argues against an erosion of statistical significance related to drop-outs. Another explanation would be a differential effect of aspirin according to time of exposure to the drug and to the natural history of polyps. The APACC study allowed distinction between the less marked and less immediate effect of aspirin on true new polyps and its effect on missed polyps. The putative anti-proliferation effect of relatively high doses of aspirin (300 mg/day) might explain the net preventive effect observed at 1 year and might be distinguished from the genuine chemopreventive effect. The latter effect might be observed at lower doses only for longer durations of treatment (at least 10 years) [10••]. The protective aspirin effect in Lynch syndrome was only observed after more than 5 years of follow-up ( and unpublished data).
Main final results of primary end points using various modes of analysisa
Numbers at randomization
Aspirin all doses
Aspirin 160 mg
Aspirin 300 mg
n = 132
n = 140
n = 73
n = 67
Analysis at 1 or 4 years
n = 116
n = 128
n = 68
n = 60
Patients with at least one adenoma, %
Adenomatous polyp burden, mm
6.3 ± 11.4
4.1 ± 6.7
3.6 ± 6
4.7 ± 7.5
Asp 160 vs Pla: P = 0.06
Asp all doses vs Pla: P = 0.07
Mean number ± SD of adenomas
1.5 ± 2.4
1 ± 1.5
0.9 ± 1.2
1.2 ± 1.7
Asp 160 vs Pla: P = 0.05.
Asp all doses vs Pla: P = 0.07
Patients with at least one advanced adenoma, %
Analysis at 4 years
n = 83
n = 102
n = 55
n = 47
Patients with at least one adenoma, %
Adenomatous polyp burden, mm
3.4 ± 6.2
3.1 ± 5.8
2.1 ± 5.4
4.3 ± 6.1
Mean number of adenomas ± SD
0.8 ± 1.3
0.7 ± 1
0.4 ± 0.7
1 ± 1.1
Asp 160 vs Pla: P = 0.025
Patients with at least one advanced adenoma, %
Comparative effects of aspirin on adenoma recurrence compared with placebo in the four published randomized controlled trials
Daily doses of aspirin, mg; MDTM
Exposure (patient-years); Recurrences under placebo
Patients with at least one adenoma
Patients with at least one advanced adenoma
Mean number ± SD of adenomas per patient
APACC 1 year [26•]
A 38/126 (30%) vs P 46/112 (41%)
A 8/126 (6%) vs P 13/112 (12%)
A 0.45 ± 0.15 vs P 0.86 ± 0.3
RR: 0.73 (0.52–1.04)
RR: 0.55 (0.24–1.27)
P = 0.08
P = 0.01
APACC 4 years
A 65/128 (51%) vs P 61/115 (53%)
A 18/128 (14%) vs P 18/115 (16%)
A 1 ± 1.5 vs P 1.5 ± 2.4
RR: 0.96 (0.75–1.22)
RR: 0.90 (0.50–1.64)
P = 0.07
A 99/434 (23%) vs P 121/419 (29%)
A 41/434 (9%) vs P 63/419 (15%)
A 0.3 = 0.7 vs P 0.5 = 0.9
RR: 0.79 (0.63–0.99)
RR: 0.63 (0.43–0.91)
P = 0.04
P = 0.013
P = 0.015
A 43/259 (17%) vs P 70/258 (27%)
A 0.3 ± 0.9 vs P 0.5 ± 1.0
RR: 0.65 (0.46–0.91)
P = 0.004
P = 0.003
140/366 (38%) A 81 mg vs 160/355 (45%) A 325 mg vs 171/363 (47%) P
28/366 (8%) A 81 mg vs 38/355 (11%) A 325 mg vs 47/363 (13%) P
RR A 81 mg: 0.81 (0.69–0.96)
RR A 81 mg: 0.59 (0.38–0.92)
RR A 325 mg: 0.96 (0.81–1.13)
P = 0.04
P < 0.05
A recent meta-analysis comparing any aspirin dose with placebo in the prevention of recurrent adenomas, and using patient-level data (3,000 patients), found a significant 17% decrease in adenoma risk and a 28% reduction in the risk of advanced lesions [34••]. The greatest benefit appeared during the first year of treatment. Contrasting with the trials of Baron et al.  and Logan et al. [22••], aspirin had no effect on recurrence of advanced adenomas in our experience, a result probably explained by the low numbers of patients with recurrent advanced adenomas. The proportions of advanced adenomas in the placebo groups were similar among the three studies assessing them.
The APACC trial found a significant effect of low-dose aspirin on adenoma prevention at 1 year, but not at 4 years. This trial was included in a meta-analysis, showing that aspirin might be used in secondary chemoprevention, in addition to colonoscopy in high-risk patients. Discrepancies between trials could be explained by differences in levels of risk for adenoma. Two major unsolved issues are the prediction of potential responders to aspirin and the choice of the best dose of aspirin and minimum duration of treatment in selected high-risk patients.
No potential conflicts of interest relevant to this article were reported.