A NON-DAIRY PROBIOTIC’S (POI) INFLUENCE ON CHANGING THE GASTROINTESTINAL TRACT’S MICROFLORA ENVIRONMENT

Abstract

Justification

Yogurt has been historically used to restore gut microflora adversely affected by antibiotic treatment. Certain fermented dairy products are probiotics; “live microorganisms which when administered in adequate amounts confer a health benefit to the host.” Microorganisms in foods may benefit certain health conditions such as diarrhea, gastroenteritis, irritable bowel syndrome, inflammatory bowel disease, and cancer. A potential new probiotic from a Polynesian traditional food is poi; a starchy paste made from the corm of taro plants.

Objective

The purpose of this study was to determine if consumption of poi, a potential non-dairy probiotic, altered the microflora in the gastrointestinal tract of healthy adults.

Methods

A cross-over clinical study included 18 subjects (19–64 years of age) divided into a poi group (n=10) and control group (n=8). The study duration of 14 weeks consisted of a 2-week washout, 4-week treatment or control, a subsequent 2-week washout, cross-over of 4-week treatment or control, and a final 2-week washout. Subjects thus served as their own controls. While receiving the poi treatment, participants consumed fresh poi (1–2 days old) three times a day (130 g/meal or about _ cup/meal); the control group did not. Both groups filled out 3-day dietary records to ensure compliance. Measurable outcomes include pre-and post-treatment microbiological fecal culture analyses.

Results

We found no significant differences in total bacterial counts following a poi diet versus following a control diet, nor were significant differences found in counts of specific bacterial species. Lactococcus tends to be higher in poi when it is analyzed for specific bacteria, but the poi consumption in our study did not alter the mean concentration of individual bacterial species (log₁₀ CFU/g wet feces) for Escherichia coli, Enterobacter, Klebsiella, Lactobacillus, Lactococcus, and Bifidobacterium. No significant differences in stool frequency or consistency were observed between the treatment and control group periods.

Conclusion

Poi consumption did not significantly alter total or individual bacterial counts in the human gastrointestinal tract. Further research might determine if “sour poi” (3–4 days old) has a greater affect than “fresh poi” (1–2 days old) as a potential probiotic, and a larger trial with longer diet durations may detect more subtle effects of poi consumption on bacterial counts.

Probiotics are defined by the World Health Organization (WHO) as “live microorganisms which when administered in adequate amounts confer a health benefit to the host.” It is believed that microflora contribute to the host’s health by improving the intestinal tract’s microbrial balance.¹ Fermented dairy products such as yogurt and buttermilk are common probiotics. The idea of purported health benefits from probiotics dates back nearly 100 years to Elie Metchnikoff who suggested that Bulgarian peasants lived longer lives because of their yogurt consumption. Certain people in Asia thought the same way and a Japanese physician, Minoru Shirota, suggested in the 1930s that the right concentration of intestinal flora could prevent disease. The Okinawans, some of the longest living people on earth, regularly consume miso soup, a traditional Japanese soup that includes miso (a thick paste made from fermented soybeans with salt).

The probiotic theory is supported by the fact that a disruption in the intestine’s delicate balance may contribute to diarrhea, gastroenteritis, constipation, irritable bowel syndrome, inflammatory bowel disease (Crohn’s disease and ulcerative colitis), food allergies, and certain cancers.^2,3 On the contrary, a balanced or “normal” enteric flora may competitively exclude possible pathogenic organisms and stimulate the intestinal immune system.^4,5

Most of the probiotic research to date has focused on dairy-based products. However, many other cultures have non-dairy foods/beverages containing microorganisms that might also serve as probiotics. Taro (Colocasia esculenta L.) is a root belonging to the Araceae family that originated in Asia and is now found primarily in tropical and subtropical regions.⁶ Historically it has been a major dietary staple in the islands of the Pacific, especially Hawaii, New Zealand, and Indonesia. Poi is the starchy paste of taro corm and primarily provides carbohydrates as well as other nutrients.⁷ Poi is made by cooking, crushing, and pounding the taro corms into a paste by adding water.⁸ The amount of water determines the thickness of poi that is then strained through a cloth.⁹ Yeast and lactic acid bacteria naturally found on the plants surface ferment the mixture eventually leading to “sour” poi (approximately 3 or more days).¹⁰ Fermentation proceeds without inoculated starter cultures and usually takes approximately two to three days. The fermentation of fresh poi is similar to that occurring in the preparation of yogurt and sauerkraut. The acid production in poi changes the pH from 6.3 to 4.5 within 24 hours and reaches its lowest pH on the fourth or fifth day of fermentation. As early as 1933, Allen and Allen¹¹ recognized that souring was the result of acid producing bacteria such as Lactobacillus and Streptococcus. These researchers identified three Lactobacillus species and two Streptococcus (recently renamed to Lactococcus). More recently Huang et al,¹⁰ identified the predominant species in sour poi as Lactococcus lactis. Unpublished reports indicate that poi made from different varieties of taro may differ slightly in the type and amount of bacteria.

The purpose of this study was to determine if poi is a potential probiotic by studying if poi consumption changes the microflora environment of the gastrointestinal tract. Fermented dairy products such as yogurt have been used historically for their action as probiotics—foods with live microorganisms that may benefit the host’s health by improving its intestinal microbrial balance. Probiotics have been reported beneficial for health problems including diarrhea, gastroenteritis, irritable bowel syndrome, inflammatory bowel disease (Crohn’s disease and ulcerative colitis), cancer, improved immune function, lactase digestion, hyperlipidemia, hepatic diseases, Helicobacter pylori infections, and genitourinary tract infections. Poi might also benefit patients through use as an allergy food substitute, and as a potential nutritional supplement for weight gain in patients with weight loss problems: failure-to-thrive, cancer cachexia, and pancreatitis/cystic fibrosis.

Studies of the potential benefits of probiotics are often designed to determine if the food changes gastrointestinal tract microflora concentration. Spanhaak et al,¹² conducted a double-blind, placebo-controlled study (20 men; 10/group) utilizing a commercially available probiotic, and reported an increase in Lactobacillus count. Benno and Mitsuoka¹³ reported that Bifidobacterium longum administered as a pharmaceutical in adults resulted in higher fecal bifidobacterial (potentially beneficial) and lower clostridial (potentially pathogenic) counts, lower fecal pH, and lower fecal ammonia concentrations. A study of 64 adults by Ling et al,¹⁴ showed that consumption of Lactobacillus GG resulted in higher fecal counts of Lactobacillus GG, decreased fecal beta-glucuronidase, nitroreductase, and glycocholic acid hydrolase activities. Ling et al,¹⁴ also showed a decrease in urinary p-cresol excretion, a product of colonic Bacteroides fragilis. Studies of infants bottle-feed an inoculum of Bifiodbacterium bifidum compared to a control showed that Bifidobacterium bifidum appeared in the stools of infants in the treatment groups.^{15, 16}

Poi has also been studied as a potential treatment for infant allergies and infant failure-to-thrive. These studies were conducted in Hawaii where some hospitals in the early 1950s were using poi on a regular basis.¹⁷ Although these older studies did not investigate poi’s probiotic potential, perhaps the observed benefits were due to its probiotic properties. In fact, based on these previous studies reviewed by Brown and Valiere,¹⁸ poi did appear to have a beneficial affect on the host’s health and therefore meets the definition of a probiotic. To date, no research has investigated the effect of poi on gastrointestinal tract microflora. While the most common bacteria in poi are Lactococcus lactis (95%) and Lactobacillus (5%) differs from that of dairy products, the two foods are similar in that the predominant bacteria in both are lactic acid producers.

Poi has some potential advantages compared to yogurt in therapeutic use in cancer patients. Although the role of an alkaline ash diet in the treatment of cancer is debatable, poi contributes to an alkaline diet since its calories are primarily derived from starch rather than sugar and protein. Also, poi is easily digestible due to its very small starch granule size, and poi is culturally acceptable to many people of Hawaiian or Polynesian ancestry.

METHODS

Subjects

Twenty-one healthy adult volunteers (aged 18–64 years) were recruited for the study. Exclusion criteria included presence of illness or disease, history of gastrointestinal, hepatic, or renal disease, obesity (Body Mass Index [BMI] > 30 kg/m²), abnormal blood pressure (WHO criteria; 140/90 mmHg), taking medication that could affect either the intestinal flora and/or the immune system and hematological and biochemical parameters, unwilling to eliminate fermented foods/beverages from diet, women who are pregnant, suspect they are pregnant, or are lactating, unwilling to consume 390 gms (375 ml) of poi per day, poi consumption within the last three months, unwilling or unable due to physicians orders to be prescription free during study, any Hawaiian ancestry (to reduce probability of any previous presence of poi bacteria), internal infection during the study, and any use of antibiotics or sulfonamides during the study. Eighteen subjects completed the study and contributed data to the analysis, while three subjects were lost due to: 1) pregnancy, 2) uncomfortable collecting stool samples, and 3) antibiotic treatment.

Recruitment

Subjects were recruited from campus at the University of Hawaii at Manoa (UHM) from among faculty, staff, and graduate students. Announcements to each of the departments were provided via flyers and emails to department heads and/or graduate program chairs to be forwarded to their faculty, staff, and graduate students. Subjects gave informed consent and the study was approved by the University of Hawaii’s IRB—the Committee on Human Studies.

Experimental Design

The study was a randomized, cross-over design involving 18 subjects (10 with a poi diet initially, 8 with a control diet initially). Duration of the study was 14 weeks consisting of a 2-week washout (regular diet without poi and following the Diet Restriction List), 4-week treatment (poi diet) or control, 2-week washout, cross-over of 4-week treatment or control (subjects on poi diet switched to control and vice-versa so that subjects acted as their own controls), followed by a final 2-week washout period. Measurable outcomes included the collection of 5 fecal samples before and after each 4- week treatment period and 2 weeks after study cessation, frequency and consistency of stool, and reported side effects.

Diet

Fresh poi (less than 24 hours old) was purchased from the HPC Foods Ltd, (Honolulu, HI) every three days. The bacterial count in fresh poi (24 hours or less old) averages one half to one million bacterial count per gram, and approximately 5 million or less in sour poi (3 or more days). Subjects were asked to keep the poi samples covered and refrigerated, which stabilized bacterial growth. The poi was also microbiologically analyzed by Brain Wave Technologies (Madison, WI) to validate bacterial concentrations.

The 4-week poi diet included poi three times a day (390 gm/day allotted into three dosages of 130 gm/125 ml or _ cup servings using volume as measurements; servings were approximated since poi has a very sticky, thick consistency). One _ cup serving each was taken during breakfast, lunch, and dinner. Subjects did not report any problems consuming these amounts of poi, which are not commonly consumed by non-Hawaiians. The control diet contained no poi. For both diets, participants were instructed to avoid fermented foods/beverages except for alcohol. A three-day dietary recall before and during the treatment/control periods verified food and nutrient intake. Subjects were requested to return the empty poi containers to ensure compliance. The poi was provided in yogurt-like containers, stored in a department refrigerator solely for the use of this study, and picked up by volunteers every three days. All groups were given a Diet Restriction List handout providing them with a list of fermented foods/beverages to avoid during the study—soy sauce, miso, sausage (salami, bologna, etc), yogurt, buttermilk, acidophilus milk, kefir, sour cream, cheese, kimchee, sauerkraut, green olives, pickles, tempeh, or any other fermented foods/beverages except alcohol. Subjects were advised to consider decreasing by 1 _ cups or servings their starchy foodstuffs to offset the increase in calories resulting from the addition of poi to their diets (392 kcals/day). Adverse reactions were self-recorded by the subjects.

Compliance Check

All subjects were contacted during and after the study to identify compliance problems. Subjects were asked if they were following the Diet Restriction List; able to consume poi at the designated times of breakfast, lunch, and dinner; remained infection-free; continued to avoid medications that would affect gut bacteria; or experiencing any side-effects. Subjects were also encouraged to discuss any questions or concerns.

Collection Of Fecal Samples

Each subject provided five fecal samples over the course of the study. Subjects were given an anaerobic pouch, generator and clip (AnaeroPack‰ System, Mitsubishi Gas Chemical America, Inc., NY) to place their fecal samples and keep them in a reduced environment. The collection system consisted of another plastic bag impermeable to oxygen, a sachet that facilitated this environment and a clip that sealed off the bag. Indicator strips in the bag determined if an anaerobic environment was maintained. Fecal samples were then stored at 40°F/4°C prior to shipping and analysis. Samples were shipped from Hawaii to North Carolina via overnight delivery service in coolers, tested for temperature upon arrival, and analyzed for their microbial profiles.

Microbial Fecal Analysis

The presence of gut microflora was determined with before and after comprehensive stool cultures for total bacteria and specifically for Escherichia coli, Enterobacter, Klebsiella, Lactobacillus, Lactococcus, and Bifidobacterium. Fecal samples (10 gm) were diluted with 90ml sterilized peptone water and were homogenized using the Stomacher 400 Lab System (Seward, Norfolk, UK) for 2 min and 100 ul of appropriate dilution was surface plated onto various selective media. Total aerobic bacterial count was enumerated on Trypiticase soy Agar (TSA), Enterobacter (coliform) were determined from eosin methylene blue agar (EMB), Escherichia coli on MacConkey agar, Staphylococcus on Briad-Parker agar, Lactobacillus on Lactobacillus MRS agar, lactic acid bacteria Lactococcus lactis on 0.5% glucose (M17-glu; Difco Laboratories, Detroit MI), Bifidobacterium on Bifidobacterium iodoacetic acid medium (BIM 25 agar), and yeast .¹⁹ Plates were incubated at 37°C for 72 hours. Bifidobacteria colonies from the BIM 25 agar were picked, gram stained, and examined under the microscope for themorphological characteristics of bifidobacteria. Fructose-6-phosphate phosphoketalase (F6PPK) and API ZYMTM (Biomerieux, Lyon, France) activity were used to confirm the identity of bifidobacteria.^20,21

Statistical Methods

Differences in the mean number of bacteria (both specific strains and total) colony-forming units per gram (CFU/gm) in different stool samples were compared using Wilcoxon sign rank tests (with similar results found using t-tests). For further exploration of interactions, multiple linear regression and Mann-Whitney tests were used. Although results are reported in the usual units of log¹⁰ CFU/gm (log¹⁰[1000*count]), analyses were performed on the original count scale of measurement. Counts smaller than 100 were not observable and were coded to have value of 100 (= log¹⁰ CFU/gm value of 5). Differences in mean stool frequency and consistency were analyzed using paired t tests, with interactions evaluated using generalized linear models. All analysis was performed using R version 1.9.0 (R Development Core Team).

RESULTS

Subject Demographics

Eighteen healthy adults (10 females and 8 males) averaging 32 years of age (+ 12.8, range = 19–64 years), including 8 Caucasians, 8 Asians, and 2 Pacific Islanders participated in the study. One subject dropped due to her need to take antibiotics for an unrelated infection, another due to pregnancy, and a third due to feeling uncomfortable with stool collections. One subject submitted 4 instead of 5 fecal collections due to time constraints. There were no reported side effects or adverse events during the duration of the study in any of the subjects.

Bacterial Populations

The number of bacteria (both specific strains and total) colony forming units per gram (CFU/gm) in the stool samples following the poi diet were similar to the CFU/gm in stool samples following the control diet: the effects of poi consumption resulted in no significant difference in mean total bacteria count or mean specific bacteria count found in stool samples directly following the poi diet compared to stool samples directly following the control diet (Table 1). Lactococcus tends to be higher in poi when it is analyzed for specific bacteria, but poi consumption did not significantly alter the mean concentration of individual bacteria (log¹⁰ CFU/g wet feces) for Bifidobacterium, Escherichia coli EC, Escherichia coli MK , Enterobacter, Klebsiella, Lactobacterum, and Lactococcus. While these analyses suggested no significant differences, we also considered interactions of gender and age with effect, and similarly found no significant differences in outcomes. Furthermore we tested for an interaction of treatment sequence with treatment effect. For several bacteria strains this interaction was significant (Table 2). In order to confirm that washout periods were sufficient, we tested for differences in mean pre-poi diet and mean pre-control diet stool sample bacterial levels, finding no significant differences. The individual values for the fecal bacteria during before and after feeding (poi or no poi) are given in Table 3. Again, there were no significant differences.

TABLE 1.

Mean Difference in Fecal Microflora Levels for Poi and Control Groups (N=18; n=8 Initial Treatment Control, n=10 Initial Treatment Poi).

Bacteria	Control Mean + SD	Poi Mean + SD	Poi Effect	P-value
Bifidobacterium	6.19 + 0.81	6.45 + 0.83	0.26	0.347
Escherichia coli EC	6.46 + 0.92	6.26 + 1.40	−0.20	0.603
Escherichia coli MK	6.24 + 0.89	6.65 + 0.40	0.41	0.644
Enterobacter	5.08 + 0.20	5.16 + 0.21	0.08	0.324
Klebsiella	5.15 + 0.34	5.26 + 0.42	0.11	0.833
Lactobacillus	6.71 + 0.59	6.95 + 0.63	0.24	0.64
Lactococcus	6.62 + 0.51	6.88 + 0.37	0.27	0.347
Total Bacterial Count	7.12 + 0.51	7.09 + 0.37	−0.03	0.468

P-value of null hypothesis that mean level after poi is different than mean level after control, Wilcox on sign rank test. Values are log₁₀ CFU/gm.

TABLE 2.

Effect of Initial Treatment of Poi or Control On Mean Levels of Fecal Microflora (N= 18; n=8 Initial Treatment Control, n=10 Initial Treatment Poi).

Strain	Initial Treatment	Control Mean + SD	Poi Mean + SD	Difference	P-value
Bifidobacterium	Control	5.74 + 0.43	7.21 + 0.46	1.48	<.01
	Poi	6.56 + 0.87	5.84 + 0.45	−0.72
Escherichia coli EC	Control	6.24 + 0.65	6.45 + 1.97	0.20	<.01
	Poi	6.63 + 1.09	6.11 + 0.80	−0.52
Escherichia coli MK	Control	6.24 + 0.74	6.80 + 0.78	0.57	0.04
	Poi	6.24 + 1.06	6.52 + 0.51	0.29
Enterobacter	Control	5.08 + 0.14	5.09 + 0.16	0.01	0.38
	Poi	5.08 + 0.25	5.21 + 0.23	0.14
Klebsiella	Control	5.20 + 0.37	5.08 + 0.21	−0.12	0.16
	Poi	5.10 + 0.32	5.40 + 0.49	0.30
Lactobacillus	Control	6.43 + 0.58	7.34 + 0.51	0.91	<.01
	Poi	6.95 + 0.51	6.64 + 0.55	−0.30
Lactococcus	Control	6.52 + 0.45	7.00 + 0.43	0.49	0.24
	Poi	6.70 + 0.57	6.79 + 0.31	0.09
Total Baterial Count	Control	7.09 + 0.61	7.34 + 0.29	0.25	0.17
	Poi	7.15 + 0.45	6.89 + 0.30	−0.26

P-value of interaction effect due to treatment sequence, that is H0: Change in mean bacterial count is equal regardless of treatment sequence. Values are log₁₀ CFU/gm.

TABLE 3.

Composition of the Fecal Microflora in Subjects Fed Poi or No Poi (N=18; n=8 Initial Treatment Control, n=10 Initial Treatment Poi).

	Poi		Control
Fecal Microflora	Mean + SD (Log10CFU/g)
	Before feeding	After feeding	Before feeding	After feeding
Bifidobacterium	5.46 + 0.70	6.45 + 0.83	5.25 + 0.28	6.19 + 0.81
Escherichia coli EC	5.57 + 0.63	6.26 + 1.40	5.83 + 0.80	6.46 + 0.92
Escherichia coli MK	5.61 + 0.62	6.65 + 0.64	5.76 + 0.69	6.24 + 0.89
Enterobacter	5.02 + 0.07	5.16 + 0.21	5.15 + 0.35	5.08 + 0.20
Klebsiella	5.05 + 0.12	5.26 + 0.42	5.20 + 0.48	5.15 + 0.34
Lactobacillus	6.11 + 0.66	6.95 + 0.63	5.85 + 0.66	6.71 + 0.59
Lactococcus	6.32 + 0.57	6.88 + 0.37	6.04 + 0.55	6.62 + 0.51
Total bacteria	6.42 + 0.58	7.09 + 0.37	6.26 + 0.76	7.12 + 0.51

Fecal Frequency/Consistency

Sixteen of the 18 subjects recorded both their stool frequency and consistency (measured on a 1–4 scale according to the following classifications: 1–diarrhea; 2–soft; 3–average; and 4–firm) on a daily basis while on either the poi diet or the control diet. We found no significant difference between control and poi diets in either measure between control and poi diets, and there was also no significant difference between the treatment and control groups. Subjects averaged 1.9 stools/day while consuming poi and 2.0 stools/day when on the control diet.

DISCUSSION

Prior to this study, no research had been conducted to investigate the effect of poi consumption on gastrointestinal tract bacterial concentration. We hypothesized that consumption of poi, with a bacterial count exceedingly higher than yogurt, would positively influence the microflora concentration of the gastrointestinal tract. However, the results of this study do not support our hypothesis.

Our results regarding total count are similar to the majority of previously published probiotic findings. Probiotic studies often report as “increasing” intestinal bacteria count, but it appears that 16 out of 18 similar studies show no significant change in “total count.” The “increase” researchers often refer to is the specific species measure within the total count; about 15 out of 18 studies reported a change (usually the probiotic bacterium). In most of these studies, the probiotic needed to be taken continuously to maintain higher levels of a given species in the intestine.²² The level of these gastrointestinal bacteria return to pretreatment levels within approximately one or more weeks after the study concludes.^12,22–24 Bougle et al,²⁵ speculated that probiotics do not appear to “colonize” the intestine, and stated that, “the great stability of the human microflora to dietary interventions is well known. It appears that the gastrointestinal tract homeostasis changes only with artificially high dosages of probiotics (in the form of supplements; amounts not normally found in foods), illness, stress, and/or medication.

The strong tendency of the gastrointestinal tract to remain relatively homeostatic is often not emphasized in probiotic studies where bacteria are frequently administered in the form of a “supplemental pill or powder,” rather than the probiotic food itself. While it is necessary to quantify bacterial counts for scientific studies, this administration method does not actually test a “probiotic.” Also, these dosages are often well above what would normally be ingested by consumers.

Even the most popular probiotic, yogurt, does not appear to increase total bacterial count, or even increase the total count of particular species found in yogurt. Nevertheless, some researchers report that this probiotic product is able to stimulate the immune system, reduce the risk of certain types of cancers, is able to suppress the activity of many undesirable fecal enzymes, and increase interferon levels.^26,27 Perhaps poi has similar potential beneficial properties that are not measured by changes in bacterial counts. In fact, recent research in vitro shows that poi inhibits the proliferation of colon cancer cells in rats, and stimulates the immune system.²⁸

The fact that many previous probiotic studies lack a control group also contributes to the difficulty of comparing probiotc studies. Our study showed increased bacterial counts in both treatment and control groups, with no overall significance as a result. Apostolou et al,²⁹ also reported a similar phenomenon in the total bacteria shifts. Hence, studies lacking a control group might erroneously report a significant increase.

We also observed considerable individual variation in bacterial counts, which has also been reported by Amann who stated, “subjects showed high inter- and intraindividual variation.”³⁰ Very few probiotic studies utilize a cross-over design where subjects serve as their own controls, but perhaps this would help reduce the probability of drawing erroneous conclusions. As a result, very few studies report the use of a washout period to compare to our results with the exception of Amann et al,³⁰ who utilized 12 days for their washout period. Despite Amann et al’s short washout 12-day period and previous studies indicating pre-treatment bacterial levels returning within about one week, perhaps a 4-week washout study would have been warranted for our study and future studies as a precaution to sufficiently separate the treatment and control groups.

Individual variations might also be influenced by ethnic differences in intestinal flora concentrations. Apostolou et al²⁹ reported an increase in bifidobacterium in healthy individuals, but not in milk-hypersensitive individuals. We did not ask our subjects if they were lactose intolerant, but 56% (10/18) of the subjects participating in our study were of Asian (44% or 8/18) or Pacific Islander (11% or 2/18) descent, a factor that may have influenced our results since a higher percentage of lactase deficiency exists in Asian and Pacific Islander populations (60–90% non-Caucasians and 6–12% Caucasians).^31,32 Although poi is a non-dairy product that does not contain any lactose and is actually hypoallergenic (casein free), we did not exclude lactose intolerant subjects from our study.

Yet another factor influencing the concentration of bacteria in the intestinal tract that may have influenced our results is how well probiotics fare through the gastrointestinal tract. Studies show that approximately 10–30% of probiotics survive, depending on a number of variables, including the type of probiotic.^33–35 Most research conducted on probiotics uses Lactobacillus from dairy-based probiotics. Perhaps non-dairy based probioitics or strains other than Lactobacillus do not show such a significant effect in changing the gastrointestinal concentration. Prior to this study, there were no reports of supplements or foods containing Lactococcus, the predominant bacterium in poi so a comparison was not possible.

Perhaps this bacterium not normally measured in the fecal samples of probiotic studies is not a “normal” inhabitant of the gastrointestinal tract and therefore subjected to above average competition from other strains. It may increase only modestly in concentration in the midst of the more dominant strains, although it is unknown if such an effect would reduce its potential benefit.

Like all probotics regardless of their strain, the bacteria in poi may have to be resistant to acid pH and biliary acids in order to colonize the intestine,² adhere to cells, have gastric acid and bile stability, produce of antimicrobial substances, and have activity against pathogenic bacteria.³⁶ We do not know if these characteristics exist with Lactococcus lactis, which may serve as yet additional influences on the presence or absence of Lactococcus lactis in the gastrointestinal tract.

The effect of poi for these bacteria appeared to be greater among subjects who first received the control diet. It is possible that participants who first received the control diet were different in some other important unmeasured covariate, although the random assignment of study subjects to initial diet certainly decreases the likelihood of such an occurrence.

To help eliminate the wide variation in methodology of probiotic studies, we suggest the criteria listed in Table 4 for future probiotic studies. Also, we suggest that perhaps “bacterial count” is a poor indicator of probiotic power. Other measurable outcomes may be more accurate indicators of gastrointestinal change that ultimately has an effect on health—breath hydrogen, fecal enzyme activity, pH, bile steroids, immune system markers (IgA, IgG, etc), gastrointestinal symptoms, and stool frequency/consistency, none of which were measured in this study. The more accurate way to test a probiotic is to determine if it “improved health,” however this is not always easy to quantify. Quantifying gastrointestinal flora following probiotic ingestion is not a direct measure of its potential benefit. Researchers also need to address the question of species specific count (since total count rarely changes) versus the beneficial effect of a probiotic. Just because species specific count goes up, does not mean health benefits will follow. It is difficult, if not impossible, to measure every single bacterium in the gastrointestinal tract and so researchers observe a small section of the scenario not to mention numerous other factors that might be contributing to a benefit that have nothing to do with a significant difference in bacterial concentration. However, probiotic research often focuses on these numbers that are easier to elucidate than the potential health benefits yielded for conditions that are often not only multifactorial in nature.

TABLE 4.

Criteria Check-list Suggested for Future Probiotic Clinical Trial Studies.

Design
[ ]	Double-blind, placebo controlled if possible
[ ]	Cross-over design with subjects serving as their own controls
[ ]	Use of drugs, alcohol, and other influencing factors controlled
[ ]	At least 1-week duration for treatment
[ ]	At least 4-week washout period before and after treatment or control
Subjects
[ ]	At least 10 per group
[ ]	Subjects age between 18–64 years
[ ]	No ethnic group variation
[ ]	Exclusion criteria Lactase deficiency No illness or disease present History of gastrointestinal, hepatic, or renal disease Obesity (BMI > 30 kg/m2) Subjects taking medication that could affect either the intestinal flora and/or the immune system and hematological and biochemical parameters Unwilling to eliminate fermented foods/beverages Women who are pregnant, suspect they are pregnant, or are lactating unwilling/unable due to physicians orders to be prescription free during study Internal infection during the study Any use of antibiotics or sulfonamides during the study
Treatment
[ ]	Natural probiotic (food/beverage) source
[ ]	If supplement used, state how much higher than probiotic food source
[ ]	Diet eliminates other fermented foods—Soy sauce, miso, sausage (salami, bologna, etc), yogurt, buttermilk, acidophilus milk, kefir, sour cream, cheese, kimchee, sauerkraut, green olives, pickles, tempeh, or other foods/beverages
[ ]	Alcohol intake recorded
Measurable Outcomes
[ ]	Total bacterial count (log10 CFU/g)
[ ]	Species specific (log10 CFU/g)
[ ]	Fecal enzyme activity—beta-glucuronidase, nitroreductase, azoreductase, 7-alpha-dehydroxylase, etc.
[ ]	Breath hydrogen
[ ]	Fecal pH
[ ]	Immune system—immunoglobulins, interferon, etc.
[ ]	Gastrointestinal symptoms—bloating, cramping, discomfort,
[ ]	Stool frequency/consistency
[ ]	Quality of life questionnaire

When we were determining microflora bacteria, several limitations existed. Identical conditions when running analyses are preferred, however, the 18 samples from subjects was spaced to permit the laboratory to run samples on a realistic schedule. The ethnic diversity in our subject population, while inclusive, may have obscured effects of poi if some subjects were lactose intolerant. Another limitation that applies to all research testing fecal samples to determine the presence of probiotic bacteria is the influence of fecal water concentration. This factor may not only vary from individual to individual, but within individuals within in the same day. Water would definitely affect bacterial concentration and yet standard fecal analysis in probiotic studies utilizes wet weight based on grams of stool, instead of dried weight. However, drying the stool would not be possible without compromising the presence of certain bacteria. Yet, another aspect that needs to addressed is that unlike other studies, we used a food and not a supplemental pill or powder with bacteria such as Lactococcus lactis in higher concentrated form to simulate the dairy-based probiotics. We feel this approach better tests the effect of a probitoic, and not the possible influence of an artificially high bacterial concentration achieved through supplements.

CONCLUSION

Our study revealed that poi consumption did not significantly alter the total or species specific bacterial counts in the human gastrointestinal tract. Given the number of factors affecting gastrointestinal tract bacterial, the potential benefits of probiotics are numerous and warrant continued research efforts in this field.¹⁸ We did observe significant increases in some bacteria following a poi diet among participants who first were assigned to the control diet, suggesting that a longer-period study may yet substantiate benefits of poi. We suggest that animal studies and clinical trials measuring the effect of poi on health are more of a direct approach to determining the therapeutic efficacy of probiotics (bacteria in foods, not supplements). Poi consumption could be specifically studied for its potential use for infant allergies, failure-to-thrive, hypercholesterolemia, and as a nutritional supplement to combat the weight loss observed in some nursing home residents, cancer patients (unpublished case studies suggest that cancer patients prefer the cool, thick consistency of poi). Future research needs to continue for poi and other probiotics (especially those of non-dairy origin) that continue to hold promise for the host’s health.