Fat, Fiber and Cancer Risk in African Americans and Rural Africans

Rates of colon cancer are much higher in African Americans (65:100,000) than in rural South Africans (<5:100,000). The higher rates are associated with higher animal protein and fat and lower fiber consumption, higher colonic secondary bile acids, lower colonic short chain fatty acid quantities and higher mucosal proliferative biomarkers of cancer risk in otherwise healthy middle aged volunteers. Here we investigate further the role of fat and fiber in this association. We performed two-week food exchanges in subjects from the same populations, where African Americans were fed a high-fiber, lowfat African-style diet, and rural Africans a high-fat low-fiber western-style diet under close supervision. In comparison to their usual diets, the food changes resulted in remarkable reciprocal changes in mucosal biomarkers of cancer risk and in aspects of the microbiota and metabolome known to affect cancer risk, best illustrated by increased saccharolytic fermentation and butyrogenesis and suppressed secondary bile acid synthesis in the African Americans.


Supplementary Tables
The macronutrient composition as % total energy for the usual and intervention diets in Africans and Americans before and after dietary change. Group mean values for dietary analysis based on 3 day recalls in the home environment on usual diets (data from 12 rural Africans and 11 African Americans) Supplemental Group mean actual dietary intakes (SDs available) during dietary switch (food given minus food returned): The table shows the analysis of the major nutrient components given each day to the 2 groups during the 14 day dietary switch, highlighting the low animal protein and fat, and high carbohydrate and fibre content of the African American intervention, and the high animal protein and fat, low fibre content of the African intervention. Note that the actual fibre intake in the African American intervention diet was higher as the content of resistant starch was not added (estimated to be an additional 9 g/day 1 , making an average total of 53g/d).   Table 5 summarizes the H&E histological staining light microscopy results from the biopsies taken from the sigmoid, transverse, and ascending colon, which show significantly higher grades of inflammation in the lamina propria, and intraepithelial lymphocyte counts in Africans, which persisted following dietary switch. The scoring system used is shown on the first table. Note that 3 sets of biopsies were taken from each subject, hence the maximum number of measurements were 20x3 for lamina propria and intraepithelial assessments. Unfortunately some samples were lost during transport, preparation or staining. Of note, the formalin preserved samples for the first 8 African participants were lost in transportation from the rural areas. Findings from the different regions of the colon showed the same pattern of differences. Left sided eosinophilic infiltration (measured only in the sigmoid colon as changes in the other regions of the colon are considered less clinically significant) was noted in 83% of Africans and 72% of African Americans. In addition to schistosoma, intestinal spirochaetosis was also identified in the biopsies of three African subjects. Measurements of short chain fatty acid and bile acid contents in colonic evacuates (mmoles/total evacuate) showing significant population differences at baseline (Mann-Whitney), followed by significant (Wilcoxon signed rank test) reciprocal changes in African Americans and rural Africans following diet switch. Total evacuate quantities were 1.50±0.13 litres in 20 African Americans and 1.63±0.10 litres in 20 Africans (p=0.17). Using these figures plus measurements of SCFA and BA concentrations in the evacuates to calculate total colonic quantities showed that, in general, faecal concentrations reflected the same changes as colonic contents. The genus-like groups that differed between Africans and African Americans at baseline. The most significantly different taxa between 20 Africans and 20 African Americans are shown (False Discovery Rate FDR<5%). The fraction of total HITChip signal is used as a proxy for relative abundance, and the average over the samples is provided for each comparison. In line with our previous observations 2 (same populations, different individuals), of the Bacteroidetes phylum, Bacteroides is the dominant genus in African Americans, while Africans are dominated by the genus Prevotella. .5 Summary of the demographic features of the 2 groups. Age and body mass index were similar for the two groups, but Africans were significantly shorter (p=0.009, Mann-Whitney) and lighter (p=0.01). Nine African Americans and seven Africans were obese with BMI >30 kg/m 2 (p=ns, Chi-squared testing).     βMe-Mal-CoA beta-Methyl-malyl-CoA γGluCys gamma-Glutamylcysteine Explanation for the abbreviations used for metabolites measured by 1 H-NMR and shown in the networks in Figure 5.

Supplementary Note 1: Additional Study Details
Study Design: Subjects were not housed and fed their usual diets for the home environment study as validation studies had been previously conducted in both communities to verify that the diets, microbiome, metabolome and mucosal biomarkers of cancer risk, i.e. epithelial proliferation measured by Ki67staining were distinct, particularly with regard to the dietary macronutrient composition, rates of colonic saccharolytic fermentation, bile acid deconjugation and mucosal proliferation 2,3 . Recruitment would have been problematic for a 4-week in-house study, because in rural Africa, healthy middle-aged members of the population conventionally look after grandchildren while their parents work, and in America it would be difficult for employed African Americans to take 4 weeks off, thus biasing selection. Recruitment: Overall, the study was well accepted by both communities and recruitment followed the estimated timelines. Only 6 African Americans who satisfied the inclusion and exclusion criteria decided not to join the study. Recruitment was easier in rural Africa, as more potential candidates were unemployed, and only one interviewed potential candidate decided not to join the study.

Inclusion criteria
1. Informed consent 2. Ages 50-65 years, inclusive 3. BMI between 20-35 Kg/m 2 : In our last study, average BMI was in the overweightmoderately obese category for both groups, i.e. 30.5(3.0) Kg/m 2 in AAs and 28.0(1.2) in Africans. To maintain representation of the sample to the general population we will limit our inclusion criteria to the weight range of the previous study, i.e. BMI 20-35, avoid severe (Class II) and very severe obesity (Class III), and balance the 2 groups.

Exclusion criteria
1. Previous GI surgery resulting disturbed gut function due to in loss of bowel or altered anatomy 2. Any form of chronic GI disease resulting in disturbed gut function, diarrhea, and malabsorption 3. Any form of acute GI disease disturbing GI function and needing current medication, e.g.
gastroenteritis, peptic ulcer disease 4. Abnormal blood tests: CBC, ESR, urea and electrolytes, LFTs 5. The detection of previously unrecognized ulceration (with depth and >0.5cm), stricture, severe inflammation, and polyps >1cm diameter during screening endoscopy 6. History of any GI malignancy 7. Present GI malignancy, previously known or detected at screening endoscopy 8. Presence of any other form of cancer or malignancy 9. Oral or IV antibiotic therapy within the last 6 weeks 10. Unable or unwilling to modify dietary intake 11. Insulin or steroid therapy that may result in altered gut function or immunity 12. Chronic non-steroidal anti-inflammatory medication, or use of short-term NSAIDS within 4 weeks of study. 13. Known HIV disease 14. Severe obesity with BMI>35kg/m 2

Supplementary Note 2: African Living Conditions "HOME ENVIRONMENT"
The environment and living conditions of the African and African American populations are quite different. The rural South Africans live in small family communities of several traditional 'pole and dagga' (wooden poles plastered with clay) thatched circular huts ('rondavels'), now being gradually replaced with more robust brick and tin roof structures. Each community has about 5 acres of land, leased from the local chief, which supports small seasonal (during the 'rainy season' November to March) vegetable gardens that grow limited supplies of corn, pumpkins, watermelons, spinach and papayas, and a variety of animals, chickens, goats, and maybe a few cattle. Cattle are considered a sign of wealth and are used for milk and only slaughtered on ceremonial occasions. Consequently, milk products are consumed, but rarely fresh: it is left outside the huts to ferment, and then consumed with relish as 'maas'. Eggs are also eaten when available, but meat in any form is scarce and generally added as flavouring rather than forming the signature component. Foods are generally boiled, not fried, and cooked in cast iron pots on open wood fires in a separate hut. Electricity is becoming more available, but is still very rudimentary and unreliable. The diet consists chiefly of 'putu', a stiff porridge made from refined commercial corn flour called 'mielie meal', with salt and vegetables added for flavouring. It is eaten communally, and forms the bulk of the 2-3 meals consumed each day. Water is usually obtained from community wells, but also from the rivers. Roads consist of rough tracks through the bush and most people have to walk between settlements and to the closest main road to catch public transport (private minicab taxis) to the towns.
Thus, there are many environmental differences that could explain the differences in disease patterns. In this study, we focus on the differences in dietary intakes, as epidemiological surveys have concluded that they most influence colon cancer risk, and they are most modifiable.
Supplementary Note 3: Colonoscopy findings: Colonoscopy was considered normal in four Americans and ten Africans. Adenomatous polyps were found and removed in nine Americans and no Africans. Hyperplastic polyps were confirmed by biopsy in eight Americans and in two Africans: four Americans had both adenomatous and hyperplastic polys. Diverticula were seen in 14 Americans and no Africans. On the other hand, endoscopic evidence of mucosal inflammation was only seen in Africans: visual evidence of mild-moderate asymptomatic patchy colitis was seen and confirmed by histology in seven Africans. In five it involved the proctosigmoid region, in one the left colon and in one other, the whole colon. In the latter two subjects, the histology revealed dense lymphocytic infiltration associated with schistosoma. In one patient with macroscopically normal appearing mucosa, a 6 cm segment of a tapeworm was seen in the transverse colon.

Potential Confounders
Effect of Obesity: Subgroup analysis did not show any distinct differences between the moderately obese and non-obese, but our study was not powered to answer this question, and the subgroup numbers are too few to closely examine the effect of obesity. It is important to note that we used each subject as his/her own control and have reported only those bacterial groups that change systematically and significantly following the dietary intervention, detected with paired statistical testing to control host factors, such as BMI and antibiotic exposure > 6weeks prior to study, that remain approximately constant within each individual during the study period. However, we confirmed that there were no significant differences (p=0.25; Wilcoxon test) in microbiota diversity in our data between obese and lean individuals (Supplementary Figure 5).
Antibiotic Exposure: From the published evidence, 6 weeks antibiotic-free was a reasonable cut-off to use, as although short-term use of antibiotics may have some long-term influence on microbiota composition, its functional significance is doubtful. The topic has been recently reviewed by Keeney et al. 4 with some studies showing complete recovery within 4 weeks and others long term losses of specific OTUs. The functional consequence of these losses was not detected, as there were no clinical manifestations, which is not surprising given the large overlap in metabolic activity between genetically distinct microbes. Again, we need to stress that the use of each subject as his/her own control should prevent this and other factors, such as the use of antibiotics in childhood, from being a significant confounder in the interpretation of our results, namely that increasing fibre increases saccharolytic fermentation and butyrogenesis whereas increasing fat increases secondary bile aid production by the microbiota. One might also argue that taking antibiotics is part of the way of life in westernized society, and this might be one of the reasons why westernized diseases, including colon cancer, are so much more common. Finally, it should be noted that the paper by Relman's group concluded that inter-individual variation was the major source of variability between samples 5 again emphasizing the importance of using each subject as his own control.

Microbial and Metabolite Stability after Colonoscopy
Although we have generally presented the results of our analyses at two points during the study, namely prior to the first colonoscopy ('home environment' sample whilst subjects consumed their usual food, and prior to the last colonoscopy after two weeks of diet exchange, we analyzed fecal samples at 6 different time points in order to discriminate diet effect from temporal variation and the potential disturbance due to bowel washout and colonoscopy: 1. Whilst on their usual diet prior to the initial colonoscopy (ED1) 2. 1 week after the first colonoscopy (HE1) 3. 2 weeks after the first colonoscopy (HE2) 4. 1 week after the diet intervention (DI1) 5. 2 weeks after the diet intervention (DI2) 6. Prior to the final colonoscopy (ED2).
Part of the reason why we took multiple samples was because of theoretical concerns we had that colonoscopy and bowel preparation might affect the microbiota (as suggested by Gorkiewicz et al 6 and Harrell et al 7 but refuted by O'Brien et al 8 ) and their metabolism. Examination of the results of our targeted qPCR measurements of microbes (functional genes) of interest and metabolites (short chain fatty acids and bile acids) shown on Figure 2, and global HITChip phylogenetic analyses shown on Figure 3 and Supplementary Figure 6, indicate that there were no significant differences between time points 1 and 3, whilst there were between 1 and 6. Evaluation across the repeat measures by Kruskal Wallis test indicated that dietary exchange had the predominant effect.