The probiotic concept evolved from a theory initially proposed by Elie Metchnikoff, who suggested that the longevity of Bulgarian peasants was due to the ingestion of fermented milk containing bacteria that eliminated toxins produced by intestinal microflora (FAO/WHO, 2001; Schrezenmeir & de Vrese, Reference Schrezenmeir and de Vrese2001). Today, probiotic therapy is based on the use of probiotic bacteria to stimulate the growth of beneficial microorganisms in the gut, instead of pathogenic bacteria, so as to enhance the body's natural defence mechanisms (Isolauri, Reference Isolauri2001; Saarela et al. Reference Saarela, Mogensen, Fonden, Matto and Mattila-Sandholm2000).
The first report of a hypocholesterolemic effect of dairy products in humans was made in the Maasai and Samburu tribes, in Africa (Mann & Spoerry, Reference Mann and Spoerry1974; Shaper et al. Reference Shaper, Jones, Jones and Kyobe1963). These authors observed a reduction in serum cholesterol after the ingestion of large amounts (4 to 8 l/d) of milk fermented by a Lactobacillus strain. In recent years, clinical trials investigating the effects of fermented milk products on serum lipids have produced conflicting results. Some studies have shown significant reductions in serum total cholesterol and LDL (Schaafsma et al. Reference Schaafsma, Meuling, van Dokkum and Bouley1998; Anderson & Gilliland, Reference Anderson and Gilliland1999; Ashar & Prajapati, Reference Ashar and Prajapati2000), while others have been inconclusive (de Roos et al. Reference de Roos, Schouten and Katan1999; Agerholm-Larsen et al. Reference Agerholm-Larsen, Raben, Haulrik, Hansen, Manders and Astrup2000; de Roos & Katan, Reference de Roos and Katan2000; Lewis & Burmeister, Reference Lewis and Burmeister2005). Xiao et al. (Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003) showed that a fermented dairy product containing Bifid. longum reduced serum cholesterol in hypercholesterolemic individuals after 4 weeks of consumption. In contrast, Kiessling et al. (Reference Kiessling, Schneider and Jahreis2002) reported that a fermented dairy product containing Lb. acidophilus 145, Bifid. longum 913 and 1% oligofructose (prebiotic) did not significantly affect total and LDL cholesterol concentrations in 29 healthy women.
Given the inconclusive nature of available data, the present work aimed to study the effect of fermented milk containing viable Lb. acidophilus and Bifid. longum on the serum lipid levels of normal to moderately hypercholesterolemic women.
Materials, Methods and Participants
Study design
This was a double-blind, placebo controlled, crossover study with a total duration of 9 weeks. It comprised 2 treatment periods of 4 weeks separated by a 1-week washout period. Thirty-four participants completed the study and were randomly assigned to 2 groups: group A (n=19) or group B (n=15). During the first treatment period, group A consumed 125 g test product 3 times daily (total of 375 g/day), whereas group B consumed the same amount of standard yoghurt (control product, containing Streptococcus thermophilus and Lb. bulgaricus). In the second period, group A consumed the standard yoghurt while group B consumed the test product. The products were taken as part of breakfast, lunch and dinner. The test product was a fermented milk containing the bacteria Lb. acidophilus 145 (viable counts of 1·4–2·1×108 CFU/g) and Bifid. longum BB536 (viable counts of 2·7×107–1·0×108 CFU/g). Determination of viable bacteria was done by standard methods (IDFGroup-E104, 1999; ISO-20128, 2006). The nutritional composition of the test and control product was similar: 4·1 g protein, 9·5 g carbohydrates (5·5 g lactose, 4·0 g sucrose), 0·1 g fat and 55 kcal per 100 g.
We tested whether the effect of fermented milk on cholesterol was dependent of initial cholesterolemia (at baseline) of the participants. Group A and B were divided into two sub-groups with total cholesterol concentrations at baseline greater or less than 190 mg/dl. In group A, there were 10 women with cholesterol values ⩽190 mg/dl (average 161 mg/dl) and 9 women with values above 190 mg/dl (average 222 mg/dl). Group B consisted of 7 women with values ⩽190 mg/dl and 8 women with values above 190 mg/dl (average values of 174 and 215 mg/dl respectively).
All participants were examined by a physician and a nutritionist. The physician checked their health status by means of physical examination, clinical history and routine laboratory tests (haemogram, lipid profile, hepatic function, renal function and urine analysis II). The nutritionist measured their weight, height and body composition by bioelectrical impedance (Maltron BF-906 Body Fat Analyser) at baseline and endpoint. The participants were advised to avoid consuming any fermented dairy products in the 2-week period prior to the study (Agerholm-Larsen et al. Reference Agerholm-Larsen, Raben, Haulrik, Hansen, Manders and Astrup2000). Further interviews with the nutritionist were carried out just before the start of the study and after week 4 and 9 (i.e. at the end of the first and second periods respectively). Participants were requested to keep a register of any changes in eating habits or physical activity and were encouraged to maintain them together with their body weight. Adherence to the tested products was reinforced and those who reported any missing intake of these were excluded. A 24 h recall questionnaire was applied to all participants in all interviews. Nutritional composition of diets was calculated using the software Microdiet v1.1 (2000). Blood lipids (total cholesterol, LDL, HDL and triglycerides) were analyzed by standard enzymatic techniques (Allain et al. Reference Allain, Poon, Chan, Richmond and Fu1974). At the end of the intervention, all patients were again weighed, measured and their body composition analyzed by bioelectrical impedance.
Participants
Participants were users of public health centres in the Madeira Autonomic Region, Portugal, who, after being informed about the aims, design and confidentiality of the study, offered to volunteer and gave their informed consent. The study was approved by the Ethics Committee of the Regional Health Service of the Madeira Autonomic Region.
The inclusion criteria for the study were: female gender; age between 18 and 65 years; body mass index (BMI) less than 30; no coronary heart disease, diabetes, hypothyroidism or hypertension; no antidyslipidemical, antifungical, immunosuppressive or antihypertensive medication; no alcoholic habits; and no pregnancy or breastfeeding.
Forty-one people were enrolled and randomly assigned to either group A or B. Seven individuals did not complete the study because they disliked the taste of the products (n=3), could not consume the amount of product required (n=2), or got ill and chose to discontinue the treatment (respiratory tract infection medicated with antibiotic, n=2).
Statistical analysis
Lipid profiles at baseline and after the first and second treatment periods were compared by analysis of variance (ANOVA). Sample distribution normality was assessed by the One-Sample Kolmogorov Smirnov test. Comparison between groups at baseline was done using the Student's independent t test or the non-parametric alternative (Mann-Whitney test). Comparisons within each group between different periods were made by Student's dependent t test. The associations between variables were investigated using Pearson or Spearman correlation coefficients. Based a cholesterol cutoff value of 190 mg/dl (De Backer et al. Reference De Backer, Ambrosioni, Borch-Johnsen, Brotons, Cifkova, Dallongeville, Ebrahim, Faergeman, Graham, Mancia, Manger Cats, Orth-Gomer, Perk, Pyorala, Rodicio, Sans, Sansoy, Sechtem, Silber, Thomsen and Wood2003), a post-hoc analysis for the plasma lipid parameters was done within each group for patients above or below this value. The software Statistical Package for Social Sciences (SPSS) version 13.0 (2004) was used in this study. Differences were considered statistically significant when P<0·05.
Results
The mean age, percentage body fat and BMI of groups A and B at baseline is presented in Table 1. All anthropometric and food intake variables were normally distributed, except for ethanol intake. No statistically significant differences were found between groups A and B for any anthropometric and food intake variable at baseline (P>0·05).
Age and anthropometric data at baseline in both groups
Values are means±standard deviations
No statistically significant differences between groups were found for any anthropometric variable at baseline
Average levels of blood lipids (mg/dl) at baseline and after each of the periods of treatment. Standard deviation values not shown for clarity purposes. Period 1 represents test product for group A and regular yoghurt for Group B. In Period 2, treatments were switched between groups
Study parameters
Table 2 presents the average plasma lipid levels in groups A and B before intervention (baseline), after period 1 and after period 2. Results show that only the HDL changes were statistically significant (P=0·015).
Total cholesterol
We found that, in the sub-group of women with a baseline cholesterol concentration greater than 190 mg/dl, the effect of period on cholesterol average values was significant (P=0·008). ANOVA was used to determine the groups and the periods in which these differences were significant. We conclude that only the individuals with more than 190 mg/dl of the group A showed significant reductions at baseline for period 1, i.e., in the period in which the test product was consumed (P=0·001). No other statistically significant differences were found for total cholesterol between any group or subgroup.
LDL
The same sequence of statistical tests were used for LDL levels and it was found that the test product had a statistically significant effect in women with an initial cholesterol level above 190 mg/dl (P=0·014). In group A, women with cholesterol concentrations above 190 mg/dl showed a statistically significant reduction in LDL, from 129 mg/dl (at baseline) to 109 mg/dl (at the end of period 1; P=0·003). However, from period 1 to period 2 the difference in LDL concentrations (109 mg/dl to 106 mg/dl) did not reach statistical significance (P=0·760). In group B, women with baseline cholesterol concentrations above 190 mg/dl showed a statistically significant reduction from period 1 to period 2 (134·5 mg/dl to 118 mg/dl). No changes were observed at the end of period. The significant differences found refer to the period in which the two groups tried the test product and represent a 16% and 12·5% decrease in groups A and B, respectively.
HDL
The differences found in HDL plasma levels were significant and occurred in both groups. In group A, a significant reduction in HDL occurred after consuming the test product (P<0·01) and a significant increase occurred after consuming the control product (P=0·003). In group B, a significant reduction was observed after the control product (P=0·001) and no significant effect (P=0·773) resulting from the test product. As for both total cholesterol and LDL, we also checked whether initial cholesterolemia of the individuals influenced HDL and we found that the results were not dependent of this factor.
Triacylglycerols
Triacylglycerols showed little variation throughout the 9 week study. ANOVA tests showed that these fluctuations did not reach statistical significance (P=0·949).
Food intake and anthropometric characteristics
Analysis of energy, protein, carbohydrate, fat and ethanol intake as well as weight, height, BMI and percentage of fat intake showed no significant differences between groups A and B either at baseline, period 1 or 2 (data not shown). We found no significant association between these variables and any of the study parameters.
Discussion
This study showed that the fermented milk containing Lb. acidophilus 145 and Bifid. longum BB536, when compared with control yoghurt (Strep. thermophilus and Lb. bulgaricus), significantly reduced the LDL levels in the individuals with initial cholesterolemia above 190 mg/dl, this reduction ranging from 12·5 to 16%.
The decreases in LDL and total cholesterol values in this study are supported by previous studies on Lb. acidophilus and on Bifid. longum, in some cases with similar methodologies (Schaafsma et al. Reference Schaafsma, Meuling, van Dokkum and Bouley1998; Anderson & Gilliland, Reference Anderson and Gilliland1999; Ashar & Prajapati, Reference Ashar and Prajapati2000; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003).
Schaafsma et al. (Reference Schaafsma, Meuling, van Dokkum and Bouley1998) showed, in a double-blind, crossover, randomized clinical trial, that fermented milk with usual yoghurt bacteria, with Lb. acidophilus (DN112.053 and DN112.0969) and with addition of fructo-oligosaccharides, caused a decrease of 4·4% in total cholesterol and 5·4% in LDL, in a sample of 30 male individuals, during 3 weeks when compared with the same amount (375 g a day) of standard yoghurt (Schaafsma et al. Reference Schaafsma, Meuling, van Dokkum and Bouley1998). Yet, it was not verified whether this effect was due to Lb. acidophilus, the fructo-oligosaccharides or both. The levels of HDL and TG remained unchanged during the study weeks.
Ashar & Prajapati (Reference Ashar and Prajapati2000) evaluated the hypocholesterolemic effect of a fermented milk with Lb. acidophilus V3, in a parallel study of 3 weeks, and concluded that the intake of 200 ml of this test product reduced the LDL by 41% and the TC by 21% in the group of individuals with initial cholesterolemia of 2·0–2·2 g/l, an effect that can be compared with that of some statins (Ashar & Prajapati, Reference Ashar and Prajapati2000). Yet, no control by placebo or by blinding was done and all four groups of the sample had a low number of individuals (between 2 and 9). The HDL did not change in any of the groups and the TG increased in the group with total cholesterol below 2·0 g/l.
In another parallel study, but controlled by placebo and blinding, a low-fat fermented milk with yoghurt bacteria and Bifid. longum BL1 was compared with a standard yoghurt in 32 male individuals. After the intake of 300 ml/day of the test product for 4 weeks, it was possible to observe a reduction in TC in half of the individuals who had an initial level above 240 mg/dl (Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003). In this same work, the effect of Bifid. longum was also tested in mice using a lyophilized powder and it was shown that TC, LDL and TG were reduced, compared with the control product.
On the other hand, there are studies that do not corroborate our results (de Roos et al. Reference de Roos, Schouten and Katan1999; Agerholm-Larsen et al. Reference Agerholm-Larsen, Raben, Haulrik, Hansen, Manders and Astrup2000; Kiessling et al. Reference Kiessling, Schneider and Jahreis2002; Lewis & Burmeister, Reference Lewis and Burmeister2005). It is important to say that the different methodologies used in these studies make the comparisons difficult and inconsistent.
The mechanism by which some bacteria may reduce plasma cholesterol is a matter of debate. In vitro studies show that the lactic acid bacteria have the capacity to assimilate and bind to cholesterol and bile acids in the intestine and that Lb. acidophilus and Bifid. longum can incorporate cholesterol in their cell membranes. This results in the inhibition of intestinal cholesterol absorption and in the decrease of its serum levels (Dambekodi & Gilliland, Reference Dambekodi and Gilliland1998; Pereira & Gibson, Reference Pereira and Gibson2002a, Reference Pereira and Gibsonb; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003; Liong & Shah, Reference Liong and Shah2005).
The increased excretion of bile acids due to bile salt deconjugation is another proposed mechanism (Schaafsma et al. Reference Schaafsma, Meuling, van Dokkum and Bouley1998; St-Onge et al. Reference St-Onge, Farnworth and Jones2000; Kimoto et al. Reference Kimoto, Ohmomo and Okamoto2002; Pereira & Gibson, Reference Pereira and Gibson2002b; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003). Klaver and van der Meer (Reference Klaver and van der Meer1993) mentioned that cholesterol coprecipitates with the deconjugated bile acids. It is known that the serum levels of cholesterol decrease whenever the intestinal reabsorption of bile acids decrease (Klaver & van der Meer, Reference Klaver and van der Meer1993; Brashears et al. Reference Brashears, Gilliland and Buck1998; Dambekodi & Gilliland, Reference Dambekodi and Gilliland1998; St-Onge et al. Reference St-Onge, Farnworth and Jones2000; Pereira & Gibson, Reference Pereira and Gibson2002b; Tabas, Reference Tabas2002; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003).
The survival capacity of bacteria in the gastrointestinal tract, namely the tolerance to both acid and bile, seems to be crucial to the in vivo effect (Walker & Gilliland, Reference Walker and Gilliland1993; Pereira & Gibson, Reference Pereira and Gibson2002a, Reference Pereira and Gibsonb; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003). This capacity has already been proved, both in animals and humans, in vivo and in vitro, for the Lb. acidophilus and for the Bifid. longum (Harrison & Peat, Reference Harrison and Peat1975; Buck & Gilliland, Reference Buck and Gilliland1994; De Rodas et al. Reference De Rodas, Gilliland and Maxwell1996; Anderson & Gilliland, Reference Anderson and Gilliland1999; Kimoto et al. Reference Kimoto, Ohmomo and Okamoto2002; Pereira & Gibson, Reference Pereira and Gibson2002a; St-Onge et al. Reference St-Onge, Farnworth, Savard, Chabot, Mafu and Jones2002; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003).
On the other hand, the Strep. thermophilus and the Lb. bulgaricus, the two most important microorganisms used in yoghurt production and that constituted the bacteria of the control product in our study, have a low tolerance to bile salts and to acid pH and a selective preference for sugars (Akalin et al. Reference Akalin, Gonc and Duzel1997; Agerholm-Larsen et al. Reference Agerholm-Larsen, Raben, Haulrik, Hansen, Manders and Astrup2000; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003).
In our study, a significant reduction of the HDL serum levels, not specific of the test product, was verified. Some investigations show the same effect (Danielson et al. Reference Danielson, Peo, Shahani, Lewis, Whalen and Amer1989; Akalin et al. Reference Akalin, Gonc and Duzel1997; De Smet et al. Reference De Smet, De Boever and Verstraete1998; Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003), and this may increase the cardiovascular risk.
The results of this study show that the test product significantly reduces the LDL in the individuals with more than 190 mg/dl of TC. There is no obvious explanation for the effect to occur only in these individuals (Xiao et al. Reference Xiao, Kondo, Takahashi, Miyaji, Oshida, Hiramatsu, Iwatsuki, Kokubo and Hosono2003). One possible explanation is the genotype of the apolipoprotein E of the participants. The influence of diet in the plasmatic lipids can be dependent on the genotype of this apoprotein (Cobb et al. Reference Cobb, Teitlebaum, Risch, Jekel and Ostfeld1992; Dreon et al. Reference Dreon, Fernstrom, Miller and Krauss1995; Dreon & Krauss, Reference Dreon and Krauss1997; Schaafsma et al. Reference Schaafsma, Meuling, van Dokkum and Bouley1998; Masson et al. Reference Masson, McNeill and Avenell2003). According to some authors, the homozygote individuals for apo E3 are more sensitive to dietary lipids than the non-homozygote ones (van Vlijmen et al. Reference van Vlijmen, van den Maagdenberg, Gijbels, van der Boom, HogenEsch, Frants, Hofker and Havekes1994; Schaafsma et al. Reference Schaafsma, Meuling, van Dokkum and Bouley1998), whereas other authors believe that the individuals with allele apo E4 are the most reactive to the diet (Lopez-Miranda et al. Reference Lopez-Miranda, Ordovas, Mata, Lichtenstein, Clevidence, Judd and Schaefer1994; Weggemans et al. Reference Weggemans, Zock, Ordovas, Pedro-Botet and Katan2001). Presently, the effects of genetic variation are still the most controversial (Weggemans et al. Reference Weggemans, Zock, Ordovas, Pedro-Botet and Katan2001; Masson et al. Reference Masson, McNeill and Avenell2003).
Results show that the product bacteria are probably responsible for these effects, but the specific action mechanisms are still to be clarified. It is speculated that the Lb. acidophilus and the Bifid. longum decrease the intestinal absorption of cholesterol both by the bile salt enzymatic deconjugation and by cholesterol assimilation and its incorporation in the cellular membrane.
The putative beneficial effects of these products upon blood lipids and cardiovascular risk may only be established with long term studies and with a more precise characterisation of the amount and type of strains used.
This study was sponsored by Lactogal Produtos Alimentares S.A., Portugal.
