Bacteriophages are bacterial viruses, evolved to infect bacterial cells. To do that, phages must use characteristic structures at cell surfaces (receptors), and to propagate they need appropriate molecular tools inside the cells. Bacteria are
prokaryotes, and their cells differ substantially from
eukaryotes, including humans or animals. For this reason, phages meet the major safety requirement: they do not infect treated individuals. Even engineered phages and induced artificial internalization of phages into mammalian cells do not result in phage propagation. Natural transcytosis of unmodified phages, that is, uptake and internal transport to the other side of a cell, which was observed in human epithelial cells, did not result in phage propagation or cell damage. Recently, however, it was reported that filamentous temperate phages of
P. aeruginosa can be endocytosed into human and murine leukocytes, resulting in transcription of the phage DNA. In turn, the product RNA triggers maladaptive innate viral pattern-recognition responses and thus inhibits the immune clearance of the bacteria. Whether this also applies to dsDNA phages like
Caudovirales has not yet been established; this is an important question to be addressed as it may affect the overall safety of phage therapy. Due to many experimental treatments in human patients conducted in past decades, and to already existing RCTs (see section: Clinical experience and randomized controlled trials), phage safety can be assessed directly. The first safety trial in healthy human volunteers for a phage was conducted by Bruttin and Brüssow in 2005. They investigated the oral administration of
Escherichia coli phage T4 and found no adverse effects of the treatment. Historical record shows that phages are safe, with mild side effects, if any. Macrophages, key cells of the innate immune system, play a central role in mediating this response. The most frequent (though still rare) adverse reactions to phage preparations found in patients were symptoms from the digestive tract, local reactions at the site of administration of a phage preparation,
superinfections, and a rise in body temperature. These reactions might have occurred because either toxins were released from bacteria destroyed by the phages—such toxin release from bacteria can also happen with antibiotic use—or due to leftover bacterial fragments or residual components from the bacterial growth medium ("food for bacteria") present in the phage treatment when unpurified preparations were used. When bacteriophages are introduced into the body, they may be recognized as foreign entities by macrophages through pattern recognition receptors (PRRs) such as Toll-like receptors (TLRs). The binding of bacteriophages to these receptors triggers macrophage activation, leading to phagocytosis (macrophages engulf and digest the bacteriophages) and cytokine production: activated macrophages produce pro-inflammatory cytokines. These cytokines can modulate the immune response but generally do not result in significant fever when phages are used appropriately. Applying bacteriophages directly to the
mucosa targets the site of infection with minimal systemic exposure, leading to a localized immune response. Injecting bacteriophages into muscle tissue introduces them to a larger number of
macrophages in the muscle and regional lymph nodes. In intravenous injection, direct introduction into the bloodstream exposes bacteriophages to macrophages throughout the body, including those in the spleen and liver. However, significant elevations in body temperature are uncommon and typically only observed in cases of rapid phage administration or high doses. Macrophages are integral to the body's immune response to bacteriophage therapy, mediating any potential immune reactions. Intravenous administration of bacteriophages is conducted under strict medical supervision, by specialists in infectious diseases within a hospital setting, due to potential adverse reactions. Adverse reactions to intravenous bacteriophage therapy may include
hypotension, i.e., a drop in blood pressure, leading to
loss of consciousness. A sudden drop (
chills) and rise (
fever) in body temperature, known as the
Jarisch–Herxheimer reaction, can occur due to the rapid
lysis of bacteria and release of
endotoxins. Rapid bacterial lysis releases endotoxins (e.g.,
lipopolysaccharides from
gram-negative bacteria) that trigger systemic inflammatory responses, including "
cytokine storms". Continuous monitoring of heart rate, blood pressure, and temperature to detect early signs of adverse reactions is done after the intravenous phage administration. Successful treatment of life-threatening infections with intravenous phage therapy has been documented. Patients have responded to therapy after one or several intravenous administrations, clearing infections that were unresponsive to conventional treatments: phages can disrupt
biofilms, which are often resistant to antibiotics, enhancing infection clearance. Bacteriophages must be produced in bacteria that are lysed (i.e., fragmented) during phage propagation. As such, phage lysates contain bacterial debris that may affect the human organism even when the phage itself is harmless. For these and other reasons, purification of bacteriophages is considered important, and phage preparations need to be assessed for their safety as a whole, particularly when phages are to be administered intravenously. This is consistent with general procedures for other drug candidates. In 2015, a group of phage therapy experts summarized the quality and safety requirements for sustainable phage therapy. Phage effects on the
human microbiome also contribute to safety issues in phage therapy. Many phages, especially temperate ones, carry genes that can affect the pathogenicity of the host. Even lambda, a temperate phage of the
E. coli K-12 laboratory strain, carries two genes that provide potential virulence benefits to the lysogenic host, one that increases intestinal adherence and the other that confers resistance to complement killing in the blood. For this reason, temperate phages are generally to be avoided as candidates for phage therapy, although in some cases, the lack of lytic phage candidates and emergency conditions may make such considerations moot. Another potential problem is generalized transduction, a term for the ability of some phages to transfer bacterial DNA from one host to another. This occurs because the systems for packaging of the phage DNA into capsids can mistakenly package host DNA instead. Indeed, with some well-characterized phages, up to 5% of the virus particles contain only bacterial DNA. Thus in a typical lysate, the entire genome of the propagating host is present in more than a million copies in every milliliter. For these reasons, it is imperative that any phage to be considered for therapeutic usage should be subjected to thorough genomic analysis and tested for the capacity for generalized transduction. As antibacterials, phages may also affect the composition of microbiomes, by infecting and killing phage-sensitive strains of bacteria. However, a major advantage of bacteriophages over antibiotics is the high specificity of bacteriophages. This specificity limits antibacterial activity to a sub-species level; typically, a phage kills only selected bacterial strains. For this reason, phages are much less likely (than antibiotics) to disturb the composition of a natural microbiome or to induce
dysbiosis. This was demonstrated in experimental studies where microbiome composition was assessed by
next-generation sequencing that revealed no important changes correlated with phage treatment in human treatments. Much of the difficulty in obtaining regulatory approval is proving to be the risks of using a self-replicating entity that has the capability to evolve. As with antibiotic therapy and other methods of countering bacterial infections,
endotoxins are released by the bacteria as they are destroyed within the patient (
Jarisch–Herxheimer reaction). This can cause symptoms of fever; in extreme cases,
toxic shock (a problem also seen with antibiotics) is possible. Janakiraman Ramachandran argues that this complication can be avoided in those types of infection where this reaction is likely to occur by using
genetically engineered bacteriophages that have had their gene responsible for producing endolysin removed. Without this gene, the host bacterium still dies but remains intact, because the lysis is disabled. On the other hand, this modification stops the exponential growth of phages, so one administered phage means at most one dead bacterial cell. Eventually, these dead cells are consumed by the normal house-cleaning duties of the
phagocytes, which utilize enzymes to break down the whole bacterium and its contents into harmless proteins, polysaccharides, and lipids.
Temperate (or
lysogenic) bacteriophages are not generally used therapeutically, since this group can act as a way for bacteria to exchange DNA. This can help spread antibiotic resistance or even, theoretically, make the bacteria pathogenic, such as in cases of
cholera. Carl Merril has claimed that harmless strains of
corynebacterium may have been converted into
C. diphtheriae that "probably killed a third of all Europeans who came to North America in the seventeenth century". Fortunately, many phages seem to be
lytic only with negligible probability of becoming lysogenic. ==Regulation and legislation==