Correspondence: Jean Paul Vernoux, Laboratoire de Microbiologie Alimentaire, Université de Caen Basse-Normandie, Esplanade de la Paix, 14032 Caen Cedex, France. Tel.: +33 0231 565621; fax:+33 0231 566179; e-mail: firstname.lastname@example.org
Lactobacilli have played a crucial role in the production of fermented products for millennia. Their probiotic effects have recently been studied and used in new products. Isolated cases of lactobacillemia have been reported in at-risk populations, but lactobacilli present an essentially negligible biological risk. We analyzed the current European guidelines for safety assessment in food/feed and conclude that they are not relevant for the Lactobacillus genus. We propose new specific guidelines, beginning by granting a ‘long-standing presumption of safety’ status to Lactobacillus genus based on its long history of safe use. Then, based on the available body of knowledge and intended use, only such tests as are useful will be necessary before attributing ‘qualified presumption of safety’ status.
Fermentation (or bioprocessing) has been used to produce a wide range of foods and food ingredients ever since the earliest recorded food preservation by humans. Many beneficial microorganisms (molds, yeast and lactic acid bacteria) are widely used to convert raw food substrates into a plethora of fermented products. Bioprocessing technology has developed this further, with specialized production of food or feed ingredients or processing aids. Bacteria of the genus Lactobacillus are beneficial microorganisms of particular interest because of their long history of use (Holzapfel, 2002). Lactobacilli were among the first organisms used by man for processing foodstuffs (Konigs et al., 2000) and for preserving food by inhibiting invasion by other microorganisms that cause foodborne illness or food spoilage (Adams, 1999). The genus Lactobacillus is essential to modern food and feed technologies, not least because of increasing interest in beneficial effects (functional properties). The dairy and self-care health industries are actively promoting the use of lactobacilli in food, and they are increasingly used in animal feed because of their potential to replace antibiotic growth promoters.
Between 1970 and 1990, the number of articles with ‘Lactobacilli or Lactobacillus’ in the title ranged between 100 and 200 per year. Most of these articles focused on their functions in food production and preservation, i.e. the traditional role of lactic acid bacteria. Interest in lactobacilli was rekindled in the 1990s following Füller's article on probiotics in 1989 (Füller, 1989) defining ‘probiotics’ as ‘live microbial feed supplements that beneficially affect the host animal by improving its intestinal microbial balance’. Since then, the beneficial effects of probiotic fermented products, many of which contain lactobacilli, have been studied extensively. The number of articles published on lactobacilli has steadily increased over the past 15 years, reaching 666 publications in 2004, 26% of which included [(lactobacilli or Lactobacillus) and probiotic] in their titles or abstracts (PubMed scientific database).
Many new products containing lactobacilli have been released on the market. They contain new Lactobacillus species of non-dairy origin, such as Lactobacillus rhamnosus GG, isolated from the healthy human intestinal flora (Goldin et al., 1992). Regulations remained unclear for some time, to the benefit of companies entering the nutraceuticals market. The industry tried to impose self-regulation as a way of avoiding the regulatory procedures required for drugs, but the need for external regulation became increasingly clear. The European Commission decided to legislate, laying down safety assessment guidelines for microorganisms present in large amounts in food or feed. Legislation began with animal feedstuffs and, in just 15 years, the feed industry has moved from a position of insufficient regulation to over-regulation. Current regulations for feed seem likely to have a direct influence on those for food, which is not so far regulated to the same degree.
Our purpose here is to review and update data on lactobacilli for food and feed purposes: the Lactobacillus genus, its long history of use, its biotope and biodiversity (focusing specifically on its probiotic use), the associated biological hazard and the European regulatory framework (focusing on the guidelines for safety assessment). Based on this synthesis, we propose a new, unified approach to Lactobacillus safety assessment for both food and feed and suggest a suitable decision tree for safety requirements.
The genus Lactobacillus
Unambiguous identification of lactobacilli is a prerequisite for risk assessment (Danan et al., 2004). Lactobacilli are gram-positive, non spore-forming rods or coccobacilli with a G+C content generally in the 33–55% range (Coenye & Vandamme, 2003). They are strictly fermentative, aerotolerant or anaerobic, aciduric or acidophilic, and have complex nutritional requirements (carbohydrates, amino acids, peptides, fatty acid esters, salts, nucleic acid derivatives, vitamins). Grown on glucose as a carbon source, lactobacilli may be homofermentative (producing more than 85% lactic acid) or heterofermentative (producing lactic acid, carbon dioxide, ethanol, and/or acetic acid in equimolar amounts).
The genus Lactobacillus was first described by Beijerinck in 1901. In 1919, Orla-Jensen divided it into three subgenera –Thermobacterium, Streptobacterium and Betabacterium– according to their optimal growth temperature and hexose fermentation pathway. The most recent edition of Bergey's Manual of Systematic Bacteriology (Kandler & Weiss, 1986), taking account of the increased number of known species, reorganizes the nomenclature into three groups: I (obligate homofermentative), II (facultative heterofermentative) and III (obligate heterofermentative). This phenotype-based nomenclature is currently accepted. A phylogeny-based nomenclature, also comprising three groups, has recently been proposed: the Lactobacillus delbrueckii group, the Lactobacillus casei/Pediococcus group and the Leuconostoc group (Collins et al., 1991; Hammes & Vogel, 1995; Vandamme et al., 1996; Stiles & Holzapfel, 1997).
Studies based on 16S rRNA genes have revealed considerable diversity in this genus. New Lactobacillus species are continually being described (http://www.bacterio.cict.fr), with 12 new species in 2005 alone. Some Lactobacillus species have been renamed: e.g. Lactobacillus bavaricus has been renamed Lactobacillus sakei; Lactobacillus cellobiosus is now Lactobacillus fermentum, Lactobacillus casei ssp. rhamnosus is now Lb. rhamnosus and Lactobacillus yamanashiensis has become Lactobacillus mali. Some former Lactobacillus species have been reclassified under new genera: Lactobacillus carnis, Lactobacillus divergens, Lactobacillus maltaromicus and Lactobacillus piscicola have been reclassified as Carnobacterium species; Lactobacillus xylosus as a Lactococcus lactis ssp. lactis; Lactobacillus minutus and Lactobacillus rimae as Atopobium species; Lactobacillus confusus, Lactobacillus kandleri, Lactobacillus minor, Lactobacillus viridescens and Lactobacillus halotolerans as Weissella species; Lactobacillus fructosus as Leuconostoc fructosum and Lactobacillus uli as Olsenella uli.
This process of taxonomic change is continuous and ongoing: 44 species and 11 subspecies were described in 1986 (Kandler & Weiss, 1986), 88 and 15 in 2003 (Coeuret et al., 2003) and 125 and 27 in December 2005 (http://www.bacterio.cict.fr). These sweeping changes cause confusion and some previous identifications of lactobacilli may yet be subject to change.
Tools for identification and characterization
The use of molecular methods has revolutionized identification, improving its quality and efficacy. The techniques used for polyphasic analysis of lactobacilli have been reviewed elsewhere (Coeuret et al., 2003). New, simple, rapid procedures using genus- and species-specific primers for accurate identification of lactobacilli have been proposed (Dubernet et al., 2002; Christensen et al., 2004; Moreira et al., 2005; Rossetti & Giraffa, 2005). The entire genomes of several lactobacilli have been sequenced (Klaenhammer et al., 2002a): e.g. Lactobacillus acidophilus NCFM; Lactobacillus brevis ATCC367; Lactobacillus casei ATCC334; Lactobacillus delbrueckii bulgaricus ATCCBAA365; Lactobacillus gasseri ATCC33323; Lactobacillus johnsonii NCC533; Lactobacillus plantarum WCFS1. Based on available data, the Lactobacillus genome varies from 1 993 564 bp for Lb. acidophilus NCFM (Altermann et al., 2005) to 3 308 274 bp for Lb. plantarum WCFS1 (Boekhorst et al., 2004). Sequencing is a very promising tool for taxonomic and functional analysis of the Lactobacillus genome.
The long history of Lactobacillus use in fermented food
Man and microorganisms are engaged in constant competition for food supplies (Steinkraus, 2002). The Earth is thought to be about 4.5 billion years old, and microorganisms were the first life-forms to appear. An examination of data from molecular and metabolic studies and paleontology led Tailliez (2001) to suggest that lactic acid bacteria, including lactobacilli, may have emerged 3 billion years ago.
Microorganisms perform the vital task of recycling organic matter in the environment. The lactic acid bacteria population expanded rapidly with the emergence of mammals 65 million years ago (Tailliez, 2001). Once humans had emerged and human populations increased, it became necessary to store food for winter or other periods when fresh food was not readily available. Stored foods provide an ideal substrate for bacteria, resulting in naturally fermented foods. Decisions as to what was edible and what was not depended on flavor, texture, aroma, and the presence or absence of toxic effects (Steinkraus, 1996). When humans started keeping livestock, around 8000 years ago (Tailliez, 2001), this selection became increasingly important and the long history of fermented milk-based foods began.
The first evidence of milk fermentation and the use of lactobacilli dates from the Ptolemaic period in Ancient Egypt, where it was depicted on stelae (Allen, 1936) and in hieroglyphics and engravings. The Egyptians kept cattle, goats, and sheep. The milk was kept in egg-shaped earthenware jars plugged with grass to protect it from insects, and was drunk shortly after milking. Due to the hot climate, in which milk is likely to spoil within a few hours, it is often assumed that milk not destined for immediate consumption was processed into something similar to quark or yogurt-like labaneh. Archaeologists have discovered that as far back as 6000 B.C. (Courtine, 1973), cheese was made from cow's and goat's milk and stored in tall jars. Egyptian tomb murals from 2000 B.C. show butter and cheese being made, and other murals show milk being stored in skin bags hung from poles, demonstrating the dairy husbandry practices of the time (http://nefertiti.iwebland.com/timelines/topics/drink.htm). From ancient Greece and Rome we have descriptions of cheese-making from such authors as Homer and Aristotle, Varro and Columella.
The Roman Legions were instrumental in spreading the art of cheese-making throughout Europe. By this time, the ripening process had been developed and it was known that various treatments and conditions during storage resulted in different flavors and characteristics. Large Roman houses had a separate cheese kitchen, the caseale, and special areas for cheese ripening. In large towns, home-made cheeses could be taken to special centers to be smoked. The changes that the Romans introduced into cheese-making are well documented (Courtine, 1973).
In the Middle Ages, cheese-making was greatly improved in the monasteries and feudal estates of Europe. Monks were great innovators, developing many of the classic cheese varieties marketed today. During the Renaissance, cheese was considered unhealthy and became unpopular, but it had been returned to favor by the 19th century, when production began to move from farm to factory (Scott, 1986).
People around the world have developed a huge variety of fermented foods from the milk of cows, goats, camels, buffalos and other animals, or from equivalent plant material such as soybean and maize. Differences in fermentation technologies and microbial environments led to different tastes, forms and textures. Fermented foods (Table 1) are often intimately bound up with civilization, customs, culture, social relationships, and in some cases beliefs or religion. Thus, the use of fermented foods reflects an intimate relationship between humans and microorganisms, and lactobacilli in particular, stretching back over 8000 years.
Despite its long history of use, however, the beneficial effects of fermented milk have been described only recently. A Russian-born biologist, Elie Metchnikoff, is well known for his study of lactic acid-producing bacteria. He attributed the long lifespan of many Bulgarians and of villagers in the Caucasus Mountains (in the area now known as Azerbaijan) to their consumption of fermented milks – yogurt and kefir (Metchnikoff, 1908). He named the main microorganism used in yogurt making Lactobacillus bulgaricus and convinced many that its subsequent colonization of the intestine helped to normalize bowel movements and combat intestinal disease, thereby increasing longevity.
This brief history highlights the ‘familiarity’ of food-fermenting microorganisms, including lactobacilli, as defined by Danan et al. (2004). It also suggests that lactobacilli should be regarded as safe, beneficial, ancestral companions of human and animals.
Biodiversity of lactobacilli
Lactobacilli are ubiquitous, being found wherever substances rich in carbohydrates are available. They are therefore closely associated with terrestrial and marine animals (Ringø & Gatesoupe, 1998; Tailliez, 2001), their environment (plants, materials of plant origin, manure) and their food (cheese, yogurt). They colonize various habitats in the bodies of humans and animals, such as the mucous membranes of the oral cavity, intestine and vagina. Lactobacilli are found in large numbers in the gastrointestinal tract soon after birth. In healthy humans, they are normally present in the oral cavity (103–107 CFU g−1), the ileum (103–107 CFU g−1), and the colon (104–108 CFU g−1), and they are the dominant microorganism in the vagina (Hill et al., 1984; Forsum et al., 2005; Merk et al., 2005). Distributions of lactobacilli in the gastrointestinal tract of humans, pigs, chickens, cattle, dogs, mice, rats and hamsters have been compared (Mitsuoka, 1992). The following lactobacilli have been found in these animal hosts: Lb. acidophilus, Lactobacillus murinus, Lactobacillus intestinalis, Lactobacillus salivarius, Lactobacillus agilis, Lactobacillus ruminis, Lactobacillus vitulinus, Lactobacillus hamsteri, Lactobacillus aviaries, Lb. casei, Lactobacillus reuteri and Lb. brevis. Lactobacilli are present in the diet and in man-made habitats such as sewage and fermenting or spoiling food. Their presence is therefore very closely entwined with the lives of animals and humans.
Biodiversity in food and feed
Fermented foods containing lactobacilli include many ancient staples, ranging from fermented milk, yogurt, cheese and wine to olives, pickles, sauerkraut, sourdough bread, fermented sausages and salami, as well as silage and recent probiotic dairy products. Table 2 provides an overview of the biodiversity of Lactobacillus species found in a range of food and feed products.
Table 2. Overview of the biodiversity of lactobacilli, naturally present in or starters added to fermented food and feed (probiotics excluded)
Some lactobacilli are present in the natural microflora of milk. These are naturally present in fermented products and are included among the non-starter lactic acid bacteria. They originate from plants or feed, the bodies of animals, and from premises in farms and cheese dairies. Initially present in small numbers (102–103 non-starter lactic acid bacteria g−1), they may reach 107–108 bacteria g−1 in cheese after a long ripening period. Non-starter lactic acid bacteria make a major contribution to flavor (Kanawjia et al., 1993; Beresford et al., 2001; Wouters et al., 2002). Lactobacilli may also be deliberately added as starters for fermented dairy foods, for technological reasons or to improve texture, flavor, taste or aroma.
One expanding application of lactobacilli is the addition of probiotic strains to various products to enhance the health of humans or animals through their favorable effects on the microflora of the gastrointestinal tract. Many definitions of probiotics have been proposed, but the one generally used comes from the FAO (FAO/WHO, 2002): ‘Live microorganisms, which when administrated in adequate amounts, confer a health benefit on the host’. ‘Live microorganisms’ here refers to a specific selected strain. The consumption of 1 × 106–1 × 1010 viable cells per day is required for beneficial effects to be seen in humans (Lee & Salminen, 1995; Dacosta, 2001; Luquet & Corrieu, 2005). This dose can be acquired by consuming marketed probiotic food or feed products, most of which contain lactobacilli (Mercenier et al., 2003).
Probiotic properties in humans and animals
The probiotic effects of lactobacilli in humans are fully documented and are strain- and dose-dependent. New prospects include probiotic therapy (Gill & Guarner, 2004), involving the use of higher doses. Clinical trials have often investigated the effects of ingesting capsules containing large amounts of probiotic (109–1010 CFU per capsule) (Kuisma et al., 2003; Kajander et al., 2005).
In the latter half of the 20th century, new concepts were developed for promoting animal health and ensuring growth performance, feed efficiency, and product quality (Zimmermann et al., 2001). Antibiotics were first added to feed to protect animals against infection, but as antibiotics also promote growth, this dual function led to widespread use as a feed additive. However, owing to safety concerns about the transmission of antibiotic resistance, the use of antibiotics in animal feed has gradually declined since 1990 and are banned altogether since January 2006. This situation led to the proposal of alternatives, such as probiotic microorganisms (Brambilla & De Filippis, 2005). Probiotics should be viable microorganisms that increase weight gain and feed conversion ratios (zootechnical properties) and decrease the incidence of diarrhea. Studies have not clearly demonstrated efficacy for zootechnical properties (Simon et al., 2001), but this may be due to the use of very different experimental setups. Some studies have reported that a growth enhancement effect is more likely in situations involving a stress of some kind (Yeo & Kim, 1997; Thomke & Elwinger 1998), as found on real farms rather than in university-based trials (Zimmerman, 1986). This assumes that health effects and zootechnical effects are closely related. Probiotic supplementation has been recommended for the treatment or prevention of various stress conditions and diseases in a number of species. Several studies have highlighted the value of lactobacilli in pigs, poultry, cattle, fish and other species (Table 3). Probiotic therapy is becoming increasingly popular in veterinary medicine, particularly for pets.
Table 3. Overview of beneficial effects of lactobacilli in animal production
Effects on local cell-mediated immunity of chickens, as shown by apparent lower levels of intestinal invasion and development of Eimeria acervulina (EA) oocysts, on the basis of higher levels of IL-2 secretion and lower levels of EA oocyst production
Significant increase (+6.2%) in milk production Used to treat alfalfa silage. Increases the concentration of acetic acid. When fed to lactating cows, it improved the aerobic stability of the ration and increased milk production
Overview of some probiotic products containing lactobacilli
Probiotic strains are strains selected for their beneficial properties and deliberately added in large amounts to food or feed.
In food, ‘probiotic’ is currently more of a concept than a defined product. Many commercial products make probiotic claims, only some of which are scientifically documented. Fermented dairy products are the most traditional source of probiotic strains of lactobacilli; however, probiotic lactobacilli have been added to cooked pork meat products, snacks, fruit juice, chocolate and chewing gum (Table 4).
Table 4. Examples of the diversity of probiotic lactobacilli in food and feed (whether or not marketed)
No cases of collective foodborne disease have been reported in healthy people or farm animals following the ingestion of food or feed containing lactobacilli. Isolated cases have occasionally occurred and, according to Harty et al. (1994) lactobacilli have pathogenic potential in elderly and immunocompromised patients, particularly those treated with broad-spectrum antibiotics.
Cannon et al. (2005) recently reviewed cases of lactobacillemia identified through a Medline search of articles published in English between 1950 and 1 July 2003. In the 241 isolated cases identified, lactobacilli were implicated mainly in endocarditis and bacteremia. Lactobacillus casei (35.7%) and Lb. rhamnosus (22.9%) were the species most commonly responsible for lactobacillemia and infective endocarditis in particular (also reported by Asahara et al., 2003), but other species were also implicated, including Lb. acidophilus, Lb. brevis, Lb. confusus, Lactobacillus curvatus, Lb. delbrueckii, Lb. fermentum, Lb. gasseri, Lactobacillus jensenii, Lb. lactis, Lactobacillus leichmannii, Lactobacillus paracasei, Lb. plantarum, and Lb. salivarius. Other authors have demonstrated that the frequency of Lactobacillus infective endocarditis is very low (Asahara et al., 2003); if a ratio of Lactobacillus infective endocarditis to total bacterial endocarditis is calculated, it generally falls between 0.05 and 0.4%, reflecting the inefficiency with which Lactobacillus attaches to intact cardiac valves (Gasser, 1994). Studies have reported that infective endocarditis may begin with transient bacteremia following dental work (Gasser, 1994), and that the microorganisms involved often originate from the patient's own buccal microflora (in which lactobacilli may be present at a concentration of 107 CFU mL−1 saliva; Adams, 1999) or diet (Zé-Zéet al., 2004). For example, Mackay et al. (1999) described the case of a 67-year-old man who suffered from Lactobacillus-associated endocarditis after having teeth extracted due to caries. This man consumed probiotic capsules containing a mixture of Lb. rhamnosus, Lb. acidophilus, and Streptococcus faecalis. As he found the capsules too large to swallow, he was in the habit of emptying their contents into his mouth, chewing them and then swallowing them with milk. Lactobacillus rhamnosus was isolated from several of his blood cultures and was indistinguishable from the strain cultured from the probiotic capsule.
In fact, it has frequently been stressed that lactobacillemia is an ultimate marker of serious or fatal underlying disease (Gasser, 1994; Saxelin et al., 1996; Husni et al., 1997; Borriello et al., 2003; Cannon et al., 2005). Thus, lactobacilli seem to display pathogenicity only in isolated cases and in exceptional circumstances, in patients with a predisposition to infection (immunocompromised or immunodeficient or parenteral access for the bacteria), where high doses of lactobacilli are ingested, mostly through daily consumption of probiotic capsules.
One limitation of these studies is that the number of cases of Lactobacillus genuinely causing lactobacillemia may have been overestimated due to false or incomplete identification. In a series of Lactobacillus bacteremia cases reported by Salminen et al. (2002), 27% of the 66 putative Lactobacillus isolates analyzed for species identification were found not to belong to the genus Lactobacillus. This demonstrates the necessity of molecular tools for identification (Lau et al., 2004).
Criteria for pathogenicity
In the last 10 years, a large number of articles have been published on the assessment of Lactobacillus safety, based on the use of tests for evaluating the pathogenic potential of recognized pathogenic microorganisms. Proposed indicators of possible pathogenicity include interaction with the normal microbiota of the host, enzymatic activity related to infective endocarditis (Harty et al., 1994), bacterial translocation, mucin degradation, and platelet aggregation. However, as lactobacilli are commensal microorganisms, a classic risk assessment approach like that used for pathogens (International Life Sciences, 1996) is not possible.
Factors such as adhesion which may lead to colonization are regarded as virulence factors in studies of pathogens. However, mucosal adhesion and other colonization factors are also essential features of most commensal organisms. Lactobacilli are well known to adhere to intestinal mucosa (Tuomola & Salminen, 1998). Indeed, most probiotic lactobacilli strains are initially selected on the basis of their ability to adhere to various mucosa models (Servin & Coconnier, 2003).
Production of toxic metabolites
Some strains of lactobacilli have an amino acid decarboxylase activity producing biogenic amines (tyramine and histamine in particular) (Suzzi & Gardini, 2003; Lucas et al., 2005). This is linked to spoilage or to long fermentation processes (e.g. cheese ripening) and depends on several parameters, including the qualitative and quantitative composition of the microflora, physical and chemical variables, the hygiene procedures adopted during production and the availability of precursors (Suzzi & Gardini, 2003). Biogenic amines may be harmful to consumers, depending on the dose and individuals (Halasz et al., 1994). No such potentially harmful compounds have been found in fermented milk prepared with probiotic lactobacilli.
Adverse metabolic activities, such as the induction of acidosis by d-lactic acid or the deconjugation of bile salts, may occur, depending on the host and the strain. Such processes may occur in the gut, particularly the colon, due to the activity of endogenous or exogenous flora (including lactobacilli) (O'Brien et al., 1999).
Lactobacilli have been shown to produce proteins or peptides that are toxic to other bacteria; these molecules are known as ‘bacteriocins’. They are nontoxic in humans and meet the requirements for food preservatives (Klaenhammer et al., 1993; Al-Hamidi, 2004).
Some authors have attempted to define virulence factors for lactobacilli. In vitro tests of pathogenic traits enabling bacteria to survive and colonize vascular surfaces have been designed. They concern platelet aggregation (Harty et al., 1993), binding to fibronectin, fibrogen and collagen, the production of enzymes that break down human glycoproteins, and synthesis of human fibrin clots (Oakey et al., 1995). None of the tested Lactobacillus strains (Lb. delbrueckii ssp. lactis UO 004 and Lb. paracasei Immunitas –Fernandez et al., 2005; Lb. acidophilus strains LA-1 and HN017; Lb. rhamnosus HN001 –Zhou et al., 2001) degraded mucin. However, other authors have demonstrated that the ability to bind mucosa strongly is not a prerequisite for the involvement of lactobacilli in bacteremia and that adherent lactobacilli do not stimulate platelet aggregation more efficiently than non-adherent strains (Kirjavainen et al., 1999). Attempts to link virulence factors, such as the prevalence of glycosidase and protease (arylamidase), to infective endocarditis have not been successful (Adams, 1999). No good predictors of pathogenic activity in vivo have been identified (Boriello et al., 2003). Furthermore, transit through the gastrointestinal tract interacts with the functional behavior of lactobacilli. Many of the genes of the non-pathogenic Lb. plantarum WCFS1 are expressed during passage through the gastrointestinal tract, enhancing its survival (Bron et al., 2004). They encode proteins involved in functions or pathways shown elsewhere to be important in pathogens during infection in vivo. Thus survival, rather than virulence, explains the importance of these genes during residence in the host (Bron et al., 2004). Searches for ‘typical’ virulence factors similar to those extensively studied for enterococci have been unsuccessful; no such factors have yet been identified in any Lactobacillus (Franz et al., 2005). The higher incidence of some species of lactobacilli in endocarditis may simply reflect the fact that they are proportionally more numerous in the main sources of infection, rather than any hypothetical virulence.
Encapsulated strains of lactobacilli have been known for many years (Hammond, 1967; Strus et al., 2001). In 1964, Sims isolated an encapsulated mutant of Lb. rhamnosus that was pathogenic in mice only after intravenous injection. A capsule is considered useful to pathogenic organisms because it increases resistance to phagocytosis. Much is known about the functional and technological properties of the capsule or slime (large amounts of heteropolysaccharides) produced by lactobacilli (De Vuyst & Degeest, 1999). These characteristics are therefore unlikely to be relevant to the virulence of lactobacilli.
Epidemiologic surveys also suggest that there is little need for concern about pathogenic potential in lactobacilli. Saxelin et al. (1996) reported that, following the release of Lb. rhamnosus GG dairy products onto the market in Finland in 1990, 3317 blood culture isolates associated with bacteremia were obtained in a 4-year prospective epidemiologic surveillance study of approximately 2.5 million people in southern Finland. None of the bacteria concerned was identical to the Lb. rhamnosus GG strain ingested in dairy products. Given the widespread consumption of Lb. rhamnosus GG in southern Finland, these epidemiologic surveillance data suggest that these bacteria have little, if any, pathogenic potential. Another survey was carried out between 1995 and 2000. This study confirmed that an increase in the probiotic use of Lb. rhamnosus GG had not led to an increase in Lactobacillus bacteremia (Salminen et al., 2002). These surveys provide further evidence that the lactobacilli ingested in food or feed are not pathogenic. All infections observed to date have concerned debilitated individuals, or immunocompromised individuals in whom a parenteral route of access was possible.
In 2000, Gasser estimated that the medical risk of endocarditis associated with lactobacilli was 0.11 cases per million people (Gasser, 2000). The risk of probiotic lactobacillemia has been defined as ‘unequivocally negligible’ (<1 case per million individuals) (Borriello et al., 2003).
We have made a rough risk estimation for France (population 50 million in 1969, increasing to 66 million today), as the French are among the leading consumers of dairy products, including cheese, yogurt and fermented milk (Luquet & Corrieu, 2005). Lactobacillus delbrueckii ssp. bulgaricus is consumed in large amounts in yogurt, as French regulations stipulate that this product must contain around 107 CFU g−1. According to our estimates, at least 90% of French fermented milks and 90% of all cheeses are likely to contain at least 107 CFU lactobacilli g−1 (Coeuret et al., 2003; Coeuret et al., 2004b). Thus, based on a mean consumption over the last 25 years of 22 kg of cheese/person/year (range: 19.9–24.6) in France and 18 kg of yogurt and other fermented milks/person/year (range: 11.8–21.9) (CNIEL values from 1980 to 2003, http://www.maison-du-lait.com), an average of 360 billion cells of lactobacilli are potentially ingested per inhabitant and per year in France. And yet only five cases of Lactobacillus-associated infection have been reported in France (Horeau et al., 1969; Dupont & Lapresle, 1977; Poty & Poty, 1979; Larvol et al., 1996; Penot et al., 1998), three of which were cases of endocarditis (searches of English and French publications from 1823 to 2005, PubMed and other medical search databases). Some Lactobacillus infections may go unreported, but they are not likely to be significant owing to their rarity and the uncertainty of some identifications (Gasser, 1994). The risk of Lactobacillus infection in France is therefore extremely low, at about one case per 10 million people over a period of more than a century. This risk is not correlated with the consumption of large numbers of lactobacilli. These findings are consistent with the absence of lactobacilli from the French government's official list of biological agents, established to control substances hazardous to health (Advisory Committee on Dangerous Pathogens, 2004).
Reid (2002), in response to an article by Sipsas et al. (2002), tried to determine how many people use probiotics containing lactobacilli. He compared the claimed sales of the manufacturer of Yakult (>9 billion bottles per year of a probiotic containing Lb. casei strain Shirota) plus those of Danone, Valio, and other manufacturers of probiotics containing lactobacilli (>20 billion doses per year) over the preceding 10 years to the small number of cases of lactobacillemia over the same period, in which as many as 200 billion doses of Lactobacillus-containing probiotic products were ingested worldwide.
Even enteral administration of lactobacilli may be beneficial in some cases, as demonstrated by the work of Thorlacius et al. (2003) on rats and that of McNaught et al. (2005) on humans. Thorlacius et al. (2003) demonstrated the attenuation of bacteremia and endotoxemia in cases of intra-abdominal infection in rats following the enteral administration of Lactobacillus strain R2LC. McNaught et al. (2005) demonstrated attenuation of systemic inflammatory response in critically ill patients following the administration of Lb. plantarum 299v. Lactobacillus may also have beneficial effects in the oral cavity, inhibiting cariogenic streptococci and Candida ssp. (Meurman 2005).
All these examples confirm that the host is a variable target and that, even in potentially at-risk populations, the response to lactobacilli depends on the immune system of the individual host.
In addition to these experiments, drugs based on viable lactobacilli have been developed and shown to be efficient and to have no side-effects (Table 5).
Vaginal gel application, vulvovaginitis, genital surgery
Legislation and safety aspects in food and feed
Current European legislation – particularly on safety assessment, as discussed below – was developed to apply to all microorganisms. In France, AFSSA (Agence Française de Sécurité Sanitaire des Aliments) has proposed general safety guidelines for assessing new or modified strains, or strains intended for a new use, for the food and feed industries (AFSSA, 2002).
Regulatory framework for food
For many years the food industry made extensive use of microorganisms in the total absence of regulations (though France and Denmark legislated early (Franz et al., 2005)), with no major problem. In France, there are specific regulations for yogurt, which is defined qualitatively and quantitatively (Coeuret et al., 2004b) and fermented milks, which are defined quantitatively only. In 2005, after a short period of reflection and open discussion (reviewed in Franz et al., 2005), the European Food Safety Authority (EFSA) adopted a generic approach to the safety assessment of microorganisms used in food/feed and the production of food/feed additives (EFSA, 2005a). According to the official text: ‘With the exception of those organisms not previously used to a significant degree in the preparation of a human food within the Community, those subjected to genetic modifications, and those used for the production of food additives, microorganisms for food use are not subjected to Community regulation … since there has been a history of presumed safe use’. It therefore follows that there is no need for safety assessment if the historical use of the microorganism can be proven. This point was recently discussed by European bodies, with a view to applying the concept of ‘Qualified Presumption of Safety’ (QPS) to a selected group of food microorganisms. It was proposed that: ‘Future applications involving a strain of microorganism falling within a QPS group would be freed from the need for further safety assessment other than any specific requirements introduced as a qualification. Microorganisms not considered suitable for QPS would remain subject to a full safety assessment’. Food manufacturers have criticized this attempt to regulate food microorganisms, as many well-established processes are based on spontaneous fermentation or the use of undefined cultures, and the QPS approach cannot be applied to such processes. It therefore poses a potential threat to traditional processes, increasing costs without necessarily increasing safety.
Regulatory framework for feed
The term ‘probiotic’ does not figure in animal feed regulations, although the idea is partly covered in the European feed additive regulations under the category of ‘zootechnical additives’.
Microorganisms have been deliberately added to animal feed since the late 1980s. They were strictly regulated in 1993, when they were brought within the scope of Council Directive 70/524/EEC of 23 November 1970 on additives in animal nutrition. Subsequent changes to European feed regulations can be attributed to a series of crises: bovine spongiform encephalitis (1996), dioxin contamination (May 1999), use of wastes for soil conditioning (June 1999), followed by the scheduled ban on the use of antibiotics as growth promoters and mounting pressure to ban genetically modified organisms. The feed additives regulations have also become increasingly stringent (Table 6), and microorganism safety assessment requirements have been made tougher. Since 2003 (Regulation 1831/2003), bacteria added to silage as inoculants have been classified as an additive, as they are expected to improve animal performance (McDonald et al., 1991). Every microbial strain used must now be assessed.
Table 6. Regulatory documents of importance relevant to the use of microorganisms in feedstuffs
1970 Council Directive 70/524/EEC of 23 November 1970 concerning additives in feedstuffs
→5 1973 1975 1984 1993 1996
Amendment 93/114/EC included microorganisms and enzymes within feed additive legislation Additives were classified in categories and defined in relation to their end use as preservative, aroma etc.
2003 Regulation (EC) N°1831/2003 of the European Parliament and of the Council of 22 September 2003 on additives for use in animal nutrition Replaces Dir 70/524/EEC
New feed additive regulation Additives classified according to functional properties: five categories were defined (technological, sensory, nutritional, zootechnical and coccidiostatic and histomonostatic)
2004 Updating of the criteria used in the assessment of bacteria for resistance to antibiotics of human or veterinary importance – EFSA-Q-2004-079 (adopted FEEDAP 25th May 2005) (http://www.efsa.eu.int)
2005 Eléments de réflexion AFSSA – révision des lignes directrices pour l'évaluation des additives – efficacité et essais de tolérance (22 February 2005) (http://www.afssa.fr)
Other relevant dates: Organizations for risk assessment and risk management:
Formation of the AFSSA (Agence Française de Sécurité Sanitaire des Aliments): 1998
Formation of the EFSA (European Food Safety Authority): 2002
The 2003 Regulation is based on the principle of a positive list, the ‘Community Register’. A dossier must be submitted to the authorities, following guidelines that prescribe detailed evaluation methods. This dossier should enable EFSA and Member States to evaluate:
1the identity of the product (composition, form) and analytical methods for its verification (Section II of the dossier);
2the efficacy of the product for each animal category (Section III of the dossier);
3the safety of the product for target species, indirect human consumers (animal products), people using the product (workers), and the environment (Section IV of the dossier).
For the time being, the antibiotic resistance profile is included in section II (identity of the product), rather than section IV (safety). EFSA (2005b) has defined an approach for evaluating antibiotic resistance in microbial strains used as feed additives. The evaluation begins with an assessment of the minimum inhibitory concentration (MIC) of well-identified strains against a wide range of antibiotics. If the strain is proven to be resistant to a specific antibiotic (MIC>thresholds defined in the EFSA paper), the applicant must determine the genetic basis of this resistance. If the resistance is intrinsic or acquired through a genomic mutation, the product is acceptable. Evaluation of the strain's antibiotic resistance profile and the transferability of any resistances is crucial. Antibiotic resistance is not, in itself, considered a hazard unless it renders the strain untreatable in cases of infection or unless it can be transferred to potential pathogens for which resistance could affect the outcome of treatment.
The guidelines used for assessing microorganisms are those drawn up by SCAN, the Scientific Committee for Animal Nutrition (SCAN, 2001). However, these carry no official weight as there is no EC directive on this subject (Becquet, 2003). Safety aspects take up most of the application dossier. These aspects were evaluated by SCAN until the expiry of its mandate early in 2003, and are now assessed by EFSA. Table 7 outlines the guidelines for the safety aspects of the dossier.
Table 7. Safety requirements for microbial feed additives according to SCAN guidelines (2001)
Tolerance test Effect on the microflora (microbial groups with safety implications)
Skin and eye irritancy Skin sensitization Toxic effects on the respiratory system Systemic toxicity
Genotoxicity (two different tests): a bacterial reverse mutation assay; an in vitro assay for clastogenicity in mammalian cells Mutagenicity: at two different somatic tissue sites in mutagenicity studies performed in vivo in mammals. Oral toxicity studies: the duration of the tests must be at least 90 days. The preferred mode of administration is by incorporation in feed, but if this is impractical, administration in drinking water or by oral gavage may be used
Most of the microorganisms selected for use in microbial additives are of gut origin, often isolated from the digestive tract of the principal target species, or derived from soil. An impact assessment would be required only if the organisms were not of gut origin and were not already ubiquitous in the environment
Thus, the regulations for feed are very detailed and the safety assessment of products containing microorganisms is particularly precise. However, the safety testing prescribed in the SCAN guidelines essentially equates microorganisms with chemical substances, which is a questionable approach.
Relevance of the 2001 SCAN safety guidelines to lactobacilli used in food and feed
The SCAN guidelines were designed to limit the risk of foodborne diseases associated with ingestion of microorganisms deliberately added to food or feed for technological or health reasons. In our view, this safety assessment is fully justified when uncommon species of other genera are put forward as candidate probiotics (Schultz et al., 2005), but is not relevant to lactobacilli. We need to develop a way of ranking the various genera of potential value for use as additives in feed or food, according to their degree of risk. For example, the genus Enterococcus has a high potential risk: it is emerging as a major cause of nosocomial infection (Weber & Gold, 2003) and Enterococcus isolates are increasingly found to carry virulence factors and high levels of transmissible antibiotic resistance (Choi et al., 2004; De Leener et al., 2005). The genus Bacillus has a moderate risk: some strains of Bacillus cereus are known to cause two kinds of foodborne illness – an emetic (vomiting) illness due to the ingestion of food containing cereulide, a toxin produced by the Bacillus, and a diarrheal infection due to the ingestion of bacterial cells/spores that produce enterotoxins in the small intestine (Hong et al., 2005). The Lactobacillus genus, by contrast, has a very low potential risk.
The tolerance test for target species of lactobacilli, in which these bacteria are administered at a dose at least 10 times the intended dose, should be maintained. On the other hand, studies of interactions with other digestive tract microflora by lactobacilli used as additives are not necessary if the manufacturer makes no specific claims. Indeed, current recommendations stipulate that such studies should be limited to enumerating coliform bacteria, enterococci and Clostridia, which do represent a safety risk. Lactobacilli are commensal bacteria of the gut and it is very difficult to isolate and selectively count the different bacterial populations from fecal samples using only ‘routinely used culture media’. Variations in one or two microbial groups may have little impact on the digestive microflora as a whole. Moreover the impact is impossible to estimate on other gut species, many of which are unknown and/or cannot be cultured.
SCAN considered that microbial additives should be regarded as respiratory sensitizers, requiring R42 notification. However, no professional disease or respiratory allergy involving lactobacilli has ever been reported, and scientific studies have even produced evidence that lactobacilli may have anti-allergenic potential (Vanderhoof & Young, 2003). The main risk to workers' respiratory systems is associated with the size of the particles of the final product which, for safety, should not be less than 10 μm. Skin and eye irritation and skin sensitization should be regarded as risks associated with the excipient. The manufacturer provides information about the excipient on the safety data sheets. There is therefore no need for further, expensive tests on laboratory animals. To guarantee safety, the use of personal protective equipment (gloves, glasses, protective clothing), which are available in most manufacturing processes, should be specified on the safety data sheet accompanying the marketed product.
The SCAN guidelines specify that tests must be carried out on the final additive product as ‘the microorganisms used in feed additives are not expected to be harmful per se. However, enzymes or microorganisms form only a small part of the complete additive, which, in most cases, will include ill-defined components originating from the fermentation or possibly from the carrier. Consequently, it is necessary to test the additive if used in food-producing animals to ensure that it does not contain mutagenic or otherwise toxic materials that may be passed on to human consumers of foods derived from animals given feedstuffs treated with these products’ (SCAN, 2001). With regard to lactobacilli, this would seem to be over-regulation, as no such ‘ill-defined’ components have been found. Although the final fermented product may include many completely undefined components, these components should be present at very low concentrations (the additive is added at a rate of 2–5 kg ton−1 feed) and are likely to be metabolized by the animals, resulting in the presence of negligible levels, far below the toxicological threshold of any toxic metabolites produced. It would therefore be more logical to check for the presence of such compounds in the food derived from the animal product rather than checking the additive itself.
Given this high dilution factor, and provided adequate tolerance tests have been performed with a dose 10-fold that intended, we see no advantage to testing the genotoxicity, mutagenicity or oral toxicity of an additive that will be ingested only indirectly. The long-standing, historically safe use of lactobacilli, which have been consumed in enormous quantities, should also dispense with the need for such toxicity tests in animals. The mutagenicity test and the 90-day rodent toxicity test should not be mandatory for products containing lactobacilli. These tests are intended to protect the consumer from harmful microbial metabolites entering the food chain via animal products. The qualitative and quantitative aspects of these metabolites are unclear. Little is known about their fate in the gut. However, using the methods cited in the SCAN guidelines (SOS-Chromotest for genotoxicity, Ames test for mutagenicity and eukaryotic cells, Comet assay for genotoxicity in Caco-2 enterocytes), Caldini et al. (2005) recently showed that strains of Lb. casei, Lb. acidophilus, Lb. rhamnosus, Lb. delbrueckii ssp. Bulgaricus, and Lb. plantarum may have antigenotoxic and antimutagenic activity. A simple precautionary principle should be applied, banning the use of probiotics containing lactobacilli in the feed of animals for 15 days before slaughter, as is currently the case for antibiotics.
No specific data on the environmental safety of the product should be required, as lactobacilli are commensal organisms in humans and animals.
Need for appropriate, unified, limited safety assessment for food and feed
Current feed safety guidelines result in extensive, potentially useless, safety testing, diverting resources from issues of genuine safety concern. They also increase demand for laboratory animal testing, which goes against the European Commission's official policy of limiting all unnecessary animal testing (Wessels et al., 2004). We therefore feel that lactobacilli should benefit from a special safety assessment procedure, which we describe below.
An important issue
Safety assessment of microorganisms in food and feed is a major concern for the European Community's Member States. A working paper issued for comment by Directorate C of the European Commission in 2003 suggested ‘a generic approach to the safety assessment of microorganisms used in feed/food and feed/food production’. The examples given include dairy lactobacilli, which could reasonably be considered for QPS (‘Qualified Presumption of Safety’) status. This open discussion led to EFSA's recent adoption of the principle of QPS status (EFSA, 2005a). The document also invited proposals for QPS microorganisms. Four groups of microorganisms have been put forward for consideration, including lactic acid bacteria.
Consideration of Lactobacillus rather than lactic acid bacteria
Lactic acid bacteria are gram-positive, anaerobic, microaerophilic or aerotolerant, catalase-negative rods or cocci. Their most important common property is the production of lactic acid as their sole, major or important product, and they are relevant to fermentation, bioprocessing, agriculture, food and therapy. This group can reasonably be defined as including Carnobacterium, Enterococcus, Lactobacillus, Lactococcus, Leuconostoc, Oenococcus, Pediococcus, Streptococcus, Tetragenococcus, Vagococcus and Weissella (Stiles & Holzapfel, 1997) (based on phylogenetic analysis, the genus Bifidobacterium is not considered to be lactic acid bacteria sensu stricto).
The problem is that some species of lactic acid bacteria are gram-positive pathogens (e.g. Enterococcus faecalis, Streptococcus equi, Streptococcus agalactiae, Streptococcus mutans, Streptococcus pyogenes, Streptococcus pneumoniae; Klaenhammer et al., 2002b). In our view, therefore, it would be difficult and risky to consider the lactic acid bacteria as a single, homogeneous group for safety assessment. We suggest that for safety assessments, Lactobacillus should be treated differently to other lactic acid bacteria.
Similar Lactobacillus safety assessments for food and feed use
Most animal production systems are designed to produce human food either directly, as meat, or through derived food products – milk, eggs, etc. The close link between humans and their farm animals and pets (shared environment, games, and licking behavior of pets) is another good reason why food and feed need not be treated differently in safety regulations. Regulations should therefore be harmonized, with regulations for animal feed no stricter than those for human food. Lactobacilli have been used for sufficiently long for us to assert with confidence that what is safe for humans is potentially safe for animals, and vice versa.
The long history of use concerns both humans and animals. It is relevant to safety because this long-term experience can replace toxicity and tolerance tests. Together with available scientific data, it justifies limited safety testing for these species. For example, certain strains for which a substantial body of scientific data is available in both animals and humans (Table 8), such as Lb. rhamnosus GG (ATCC 53103) and Lb. casei Shirota, should be authorized for use in animal feed following only limited testing.
Table 8. Examples of lactobacilli studied for use in animal production and for human consumption
By August 2005, the European Community had given unlimited authorization for the use of nine microbial feed additives with 22 applications in various animal species. Only one contains a strain from the genus Lactobacillus. This is Lb. rhamnosus strain DSM 7133, in association with a strain of E. faecium. Three of the 14 provisional authorizations concern products including Lactobacillus strains: Lb. acidophilus strains CECT 4529 and DSM 13241 and Lb. reuteri 1063S. Because of the very strict regulations in force, only a few feed additives containing lactobacilli are currently available in Europe. Furthermore, European feed additives generally include only one or two strains, whereas those used in the USA often contain multiple strains and multiple species. Timmerman et al. (2004) reported that multistrain and multispecies probiotics were more effective and efficient than monostrain products. In contrast, in human foods, for which there is currently no regulatory framework, about 85% of the fermented probiotic products on the market contain Lactobacillus (calculation based on the data reported by Mercenier et al., 2003). This suggests that over-regulation in the area of animal feed has greatly hampered innovation, since applying for authorization places a heavy burden on companies in terms of financial and human resources.
Proposal for a decision tree
Our proposed assessment procedure includes several steps. The first is the attribution of ‘Long-standing Presumption of Safety’ status to a taxonomic unit recognized as a companion organism in humans and animals with a long history of safe use in food and feed. As with the present system, the attribution of Qualified Presumption of Safety (QPS) status depends on the accumulation of a sufficiently large ‘body of knowledge’ (defined by EFSA, 2005a) for a genus, species or strain and on specific safety assessment. We consider that in this body of knowledge, ‘history of use’ and ‘industrial applications’ should be given greater weight than ‘ecology’, ‘scientific literature and databases’ and ‘clinical aspects’. In our view, an ‘insufficient body of knowledge’ implies that there is no historical use of the strain and no known industrial application. We suggest that 10 years is a sufficiently long period to constitute a documented history of use.
In light of the arguments developed in this article, the genus Lactobacillus should be accorded ‘Long-standing Presumption of Safety’ status. Then, depending on the body of knowledge about a particular species and its intended use, the Lactobacillus concerned would have to undergo single-criterion, dual-criterion or full safety assessment, to obtain QPS status.
Single-criterion safety assessment
This single-criterion safety assessment is relevant to all lactobacilli strains, including those with a documented history of use (Mogensen et al., 2002), the criterion in question being transferable antibiotic resistance.
There is a consensus that the antibiotic resistance of lactobacilli is a relevant and topical issue for safety assessment (Franz et al., 2005). Lactobacilli for use as probiotics should not contain transferable genes conferring resistance to antibiotics of importance in clinical human and veterinary care (Tynkkynen et al., 1998; Klein et al., 2000). Some species of lactobacilli are known to be naturally polyresistant (Franz et al., 2005). This is the case for Lb. rhamnosus GG, which is resistant to 18 antibiotics (Coppola et al., 2005). This resistance is conferred by genes not borne on plasmids (Tynkkynen et al., 1998). This single safety requirement would be a key point in the attribution of QPS status and would apply even when a sufficient body of knowledge about potential safety is available for the species in question.
Dual-criterion safety assessment
Where it is intended to use a known Lactobacillus on a new target, as well as the antibiotic resistance assessment, tolerance tests would be carried out at the intended dose and 10 times the intended dose (SCAN, 2001). This ‘dual-criterion’ requirement should ensure the safety of the final product (not only the microorganism), and also safety in the event of accidental overdose. This test should be conducted on the target animal species or, for food applications, on relevant laboratory animals.
Full safety assessment
In the absence of a sufficient body of knowledge or if a food use is envisaged for a species so far used only in animals, we propose specific tests in addition to those for antibiotic resistance and tolerance, to address the potential risk of infective endocarditis or bacteremia in human populations at risk (e.g. immunocompromised subjects, neonates, hospital patients). These specific tests correspond to the in vitro and in vivo tests proposed by Asahara et al. (2003). The in vitro test evaluates the resistance of lactobacilli strains to intracellular killing by mouse-elicited macrophages. The in vivo test involves assessing possible infective endocarditis induction in a rabbit model. In that work, positive results were obtained in rabbits for the strain Lb. rhamnosus GG and negative results for the strain Lb. casei Shirota, in contrast to previous studies in humans that had implicated both species equally in endocarditis.
If any risks are identified, we propose the attribution of an ‘experimental presumption of safety’ (EPS) status for 10 years, giving enough time for the constitution of a relevant body of knowledge for the attribution of QPS status. This 10-year period corresponds to about 10% of human life expectancy, similar to the time-scale used in tests of subchronic toxicity in animal models. Over this period an epidemiologic survey would be conducted with the intended target (animal or human).
If specific possible risks are identified, the most realistic approach would be to attribute EPSr (restricted) status for 10 years, with the following restrictions on use: (1) prohibition of ‘super-dose’ products to avoid medical or veterinary use as a probiotic therapeutic agent; (2) prohibition of the sale of capsules (to make excess consumption more difficult); (3) prohibition of the use of any food or feed product containing this strain in a medical or veterinary environment; and (4) the final ingested product (feed or food) should not exceed the current ‘normal’ doses for feed or food probiotics (106 and 107 CFU g−1, respectively; Coeuret et al., 2004b). If, after these recommendations have been followed under EPSr status for 10 years, the body of knowledge has increased in a satisfactory manner, the strain would be granted QPS status with the same restrictions on use (QPSr (restricted)).
It would have been possible to attribute non-QPS status directly to any strains giving positive results in specific tests. However, this option is not consistent with the estimated risk of around one per 10 million. As Wessels et al. (2004) point out, ‘a regulation for reasons of safety should be proportional to perceived risks, risk being a function not only of severity but also of probability of the adverse effect taking place’. The toxicologic impact of a substance is known to depend on dose, host and other associated factors (e.g. age, environment, health status). Regulations should take the entire population into account and not remove the possibility for most of the population to benefit from Lactobacillus use because a very small proportion of the population is at risk. Restrictions on use are therefore preferable to a total ban. This general policy already applies to food. It should also be borne in mind that the concept of toxicologic threshold has been applied to many contaminating microorganisms. For example, Escherichia coli, which presents a far greater risk than lactobacilli, is permitted to reach 105 cells g−1 of cheese (French regulations, Lamy Dehove, May 2002).
We therefore consider that provisional EPSr status for 10 years, followed by QPSr status with restrictions on use, would be preferable to non-QPS status.
For example, let us suppose this decision tree is applied to Lb. rhamnosus GG when it is an unidentified strain without a sufficient body of knowledge. Positive results in rabbit-specific tests (Asahara et al., 2003) would lead to an EPSr status attribution with restrictions on use, preventing possible illness caused by Lb. rhamnosus GG in some people, without depriving most people of its general health benefits. The 10-year epidemiologic survey would reveal the very negligible impact of Lb. rhamnosus GG. This scenario demonstrates the value of the EPSr step proposed in the decision tree. The debate remains open on the alternative options of EPSr and non-QPS status, but we would tend to opt for EPSr.
Our proposed decision tree for safety assessment of lactobacilli used in food or feed, based on this reasoning, is shown in Fig. 2.
Lactobacilli have clearly provided scientists (microbiologists and food/feed researchers) with many exciting research topics. Biomedical research currently offers new prospects for applications and this may have far-reaching results in the future if the international community takes up the suggestion of Reid et al. (2005) that it consider the potential benefits of lactic acid bacteria probiotics and food or dietary supplements (lactobacilli included) in slowing morbidity and mortality associated with HIV/AIDS and gastroenteritis. This approach, built upon sound clinical findings and scientific investigations, is of great potential value for remote communities with limited access to health care (Reid et al., 2005).
Our proposal for Lactobacillus safety assessment ensues from a reasonable and realistic approach to risk-benefit evaluation. We propose new, specific safety guidelines for attributing ‘qualified presumption of safety’ status according to the available ‘body of knowledge’ on each strain. This decision tree should also be relevant to any taxonomic unit in which safety is presumed, such as the genus Bifidobacterium. The main limitation of our proposal is the potential use of specific pathogenicity tests for at-risk populations. This is open to debate because, although results correlate well with clinical reports of pathogenicity, these tests relate to an almost non-existent risk – about 1 in 107, equivalent to no more than one case per million potential consumers. Including it in a decision tree might lead to a false perception of the real risk associated with the use of lactobacilli.
We therefore defend the attribution of ‘Long-standing Presumption of Safety’ status to lactobacilli and provide a realistic proposal for a unified safety assessment for feed and food use, leading to the attribution – or not – of ‘Qualified Presumption of Safety’ status. This approach should remove obstacles to innovation due to current over- and under-regulation. It should reconcile scientists and industrialists with the regulatory authorities, and help to ensure that ‘lactobacilli remain the fond favorites of many microbiologists’ (Tannock, 2004) and consumers.