Commentary on Colon Cancer Prevention Database
The chemoprevention database is available on two servers:
NACRe INRA , recommended, and Mirror site on Free.Fr with more frequent updates - Cliquer pour base de données en français
Back to main page. Here are explained the results. Other pages: methods, and discussion including ACF MDF BCAC discussion.

Tumor Free Rats: Efficacy of Chemopreventive Agents
against Experimental Neoplasms in the Gut of Rodents.

What are the most potent agents?

rat The companion database proposes a ranking of the most potent chemopreventive agents detected in rodents' studies (see chemoprevention) The present page comments data in these tables: tumor table, and ACF table (see what ACF are in the main page). The ACF table shows the efficacy of chemopreventive agents on ACF endpoints. The agents were ranked according to their potency to reduce the number of ACF per rat. Most potent agents are thus placed on the top of the table, and eleven articles report agents that reduce the ACF number by more than 80% (potency higher than 5).
      The most potent agents against ACF were
- PEG-like pluronic F68 (potency 76, i.e., pluronic treatment reduced 76-fold the ACF number),
- PEG 8000 (potencies of 56, 18, 14, 8 and 5.5 in five independent articles),
- perilla oil associated with beta-carotene (potency 11),
- indole-carbinol (potency 11 on imidazopyridine-induced ACF, but control rats had only 3 ACF/rat),
- a NO-releasing aspirin derivative (potency 7 on TNBS-DMH-induced ACF),
- sulindac sulfide (potency 7), and a
- caffeic acid ester (potency 5.6).
Among the 186 agents in the ACF table, the median agent halved the number of ACF (median potency, 2). Indole-carbinol, PEG (three studies), chlorophillin, cork, pluronic, sanshishi, auraptene (two studies), troglitazone, zerumbone, ursodeoxycholic acid and hesperidin were the most potent agents to reduce the ACF size (or crypt multiplicity), a marker of the ACF growth rate. As shown in databases, each agent was attributed to a class (last column). The mean potencies of all agents in each class on ACF were calculated, compared by ANOVA analysis, and shown on figure 1 as percent inhibitions (hatched bars). PEG class was significantly more potent than any other class (ANOVA p<0.0001). No significant difference was seen among the eight other classes (PEG-omitted ANOVA p=0.28).

Tumor table shows the efficacy of chemopreventive agents on tumor endpoints. The agents were ranked according to their potency to reduce the incidence of tumors in the colon and rectum. Sixteen articles report agents that reduce the tumor incidence by more than 80% (potency higher than 5).
      The most potent agents against tumors were
- celecoxib (potency 15),
- piroxicam or aspirin with DFMO (potencies 9 or 5.3),
- PEG (potencies of 8.6 and 7 in two independent articles),
- S-methyl methane thiosulfonate (MMTS, potency 7.9),
- Bifidobacterium longum (potency higher than 7 against imidazoquinoline initiation),
- a protease inhibitor (potency 10.4 and estimated 7.3 in two mice's studies),
- folic acid (potency 7),
- piroxicam (potency 6.5),
- pectin (potency 5.7),
- obacunone (potency 5.5), and
- magnesium hydroxide (potency 5.3).
Most of these agents were included in the diet during both initiation and promotion phases, and only PEG, MMTS, piroxicam, and obacunone were effective after the end of the initiation period (protocol labeled "post" in table 2). Some potencies were not accurately established, because the tumor incidence was small in the control group. It was often the case for the most potent agents. Indeed, potency of B. longum, protease inhibitor, folic acid, and calcium glucarate, were based on only 7 tumor-bearing rats or mice in their respective control group (see Tumor table, coumn with raw data of incidences).

Among the 160 agents in Tumor table, the median agent halved the tumor incidence (median potency, 2). No carcinoma was detected (100% inhibition of cancer) in rats fed ursodeoxycholic, PEG (two independent studies) or MMTS, and in rats given exercise. In addition, celecoxib, acetoxychavicol, selenium, p53 vaccination, piroxicam with DFMO, cellulose, aspirin, S-allylcysteine, obacunone, sulindac sulfone and hesperidin (two studies) reduced the incidence of adenocarcinoma more than 78%. Each agent was attributed to one of nine classes, as in ACF table. The mean potency of agents in each class were calculated, compared by ANOVA analysis, and shown on fig. 1 (not schown on the website: see the publication) as percent inhibition of the tumor incidence (solid bars). The mean potency of NSAIDs, PEGs, and amine modulators were not different, but PEG class was significantly more potent than the six other classes (ANOVA p=0.0004). When PEG was omitted, the ANOVA remained significant (p=0.02), but NSAIDs and amine modulators were not significantly more potent than any of the other classes (all pairwise p were larger than 0.05).

In an attempt to combine results from both tables, we merged the six tabulated endpoints in a non-parametric way. Tables were ranked sequentially according to potency of each agent to reduce
(i) the number of ACF,
(ii) the number of large ACF,
(iii) the number of crypts per ACF,
(iv) the tumor incidence,
(v) the adenocarcinoma incidence, and
(vi) the tumor multiplicity.
Obviously, these endpoints are partly redundant, but they do not fully overlap. We gathered the top-twenty agents against each endpoint in a list of 120 items (not shown).
      Most potent agents cited most often in this list were then
- PEG 8000 (cited 20 times),
- DFMO alone or with piroxicam or aspirin (cited 8 times),
- a protease inhibitor (cited 5 times),
- celecoxib (cited 4 times),
- hesperidin (cited 4 times),
- sulindac sulfone or sulfide, (cited 4 times)and
- Bifidobacterium (cited 4 times).
Other agents appeared 3 times or less in the table: they may be potent agents too, but have been reported in few articles yet. Making the list with top-12 agents instead of top-20 did not change much the most-cited agent list.

ACF-Tumor Correlation

Fifty-seven agents were found in both ACF and tumor tables.
A significant correlation was found between the potencies in the ACF assay and in the tumor assay (r=0.45, N=57, p<0.001). Celecoxib is very potent against tumors, but not against ACF. It thus appeared as a major outlier in the tumor-ACF correlation. When this outlying point was dropped, correlation increased to r=0.68 (N=56, p<0.001).
The faster ACF grow, the larger they become. Thus the ACF size may relate more closely to the tumor endpoint than the ACF number. However, many articles do not report ACF size or large ACF number, and correlation could be calculated from fewer points than above.
Correlation of the tumor incidence with ACF multiplicity (size) was r=0.69 (p=0.005, N=20), and
Correlation of the tumor incidence with the number of large ACF was r=0.76 (p<0.001, N=36). Both values are higher than the correlation with ACF numbers, which tend to support the hypothesis that ACF size or large ACF numbers are more pertinent markers of promotion than ACF numbers.
Above are explained the results. Back to main page (indroduction). Next pages: methods, and discussion

Corpet DE & Taché S, 2002, Nutrition & Cancer.