Predicting the future is always difficult, indeed impossible. We can have some theories, but many are proved wrong. Possibly one of the best ways of looking into the future is to start with history. I first became involved with the cephalosporins some 35 years ago. While working in Edinburgh, Scotland, in 1964, I had the privilege of being involved in the first clinical trial of cephaloridine (the first cephalosporin, produced by Glaxo) in the treatment of serious infections. We treated a number of patients suffering from pneumococcal meningitis and I believe that I was the first person ever to give cephaloridine by the intrathecal route. We did not know the intrathecal dose. I was asked to give 100 mg intrathecally and the patient had an unpleasant reaction, with confusion, hallucinations and nystagmus. The dose was clearly excessive and my chief at that time suggested halving the dose to 50 mg. There was no reaction, and for many years in the textbooks, the intrathecal dose of cephaloridine was quoted as 50 mg. Since then the use of cephalosporins in bacterial meningitis virtually stopped. Indeed, some publications suggested that cephalothin was ineffective in bacterial meningitis, and was potentially dangerous. However, now in the 1990s, drugs such as ceftriaxone and cefotaxime are the antibiotics of first choice for bacterial meningitis—even for pneumococcal meningitis. So, there is an old indication that disappeared but has come back again—the use of the cephalosporins in pyogenic meningitis. Next, I would like to discuss the controversy between the use of cephalosporins or aminoglycosides in the treatment of sick patients with unexplained sepsis, febrile neutropenics, etc. Gentamicin became available at approximately the same time as the early parenteral cephalosporins and, for many years, clinicians felt that the aminoglycosides, because they were more bactericidal, were better drugs than cephalosporins for sick patients with undiagnosed sepsis. However, third-generation cephalosporins are now used for the indications previously favored for gentamicin. Nevertheless, gentamicin is still in the guidelines for bacterial endocarditis and the cephalosporins still have not found a definite place for this indication. The UK recommendations for most types of endocarditis are to use an aminoglycoside plus a beta-lactam, which is usually not a cephalosporin. Possibly in the future some of the newer cephalosporins may replace the aminoglycosides for this indication. In 1964, my boss at the Hôpital Claude Bernard said: ‘cephalothin combines the properties of ampicillin and the quality of oxacillin: every time you have to prescribe one or the other of those, you can prescribe cephalothin’. Thus, cephalosporins were regarded as broad-spectrum antibiotics and able to replace penicillin against Staph. aureus (we had no methicillin-resistant Staphylococcus aureus (MRSA) at that time) and superior to ampicillin, especially against Klebsiella. Then there was a cascade of one compound after another. No class of antibiotics has been so diverse, in terms of in vitro activity, bactericidal activity, pharmacological indication, etc., making it very difficult to say that first-generation compounds can be replaced by the second-generation, etc. In the first generation, cephaloridine was less stable to beta-lactamase than cephalothin; the latter was more satisfactory for Gram-positives, and cephaloridine better against Gram-negatives. Third-generation compounds, such as cefotaxime and ceftazidime, are completely different again, which makes things extremely difficult in the laboratory. When we want to have the profile of susceptibility of a particular microorganism, how many cephalosporins do we put on the plate? Even five may not be enough: there may be a Pseudomonas aeruginosa resistant to ceftazidime and susceptible to cefepime. The situation is made worse now by the presence of extended-spectrum beta-lactamases (ESBLs). We know cephalosporins are substrates for ESBLs, so we assume that it is better not to use a cephalo-sporin in those cases; but we are only considering the activity. Classification as first, second, third generation, etc. is also difficult for oral compounds. Oral cephalosporins were primarily designed to treat respiratory infections, while the third generation was designed to confer stability on beta-lactamases of Gram-negative bacteria, but was less active on Gram-positive organisms. In 1964, we viewed cephalothin as a broad-spectrum antibiotic but we now see it as being of particularly narrow spectrum. How can we call an antibiotic broad spectrum that is not very good on streptococci, not active on enterococci or MRSA, and has limited activity on anaerobes? Thus, they are narrow spectrum, not very different from an aminoglycoside. I think cephalosporins should be referred to by their names and not by their generation. Each compound has a certain type of indication and has to be reappraised in certain conditions, especially for the dosage and the way it is used. Finally, I feel that oral cephalosporins are usually underdosed. There should be a reappraisal of their pharmacodynamics. The idea has been developed throughout this conference of ‘therapeutic niches’ rather than general cephalosporins. In 1964, similarly to Alastair Geddes, I was working on cephaloridine in the laboratory. Cephaloridine was introduced as a niche compound, for the treatment of staphylococcal infection. However, our work suggested that staphylococcal beta-lactamase hydrolyzed cephaloridine at a rate that could be clinically significant. Therefore, my credential for being here is that, in 1964, I was denying that our first cephalosporin had any place in the therapeutic armamentarium! In fact, it was very popular to do that in the UK for quite a number of years, and some very distinguished microbiologists said that, if cephalosporins disappeared, nobody would be in the least bit worried. We have moved on since then. I wish to ‘nail my colors to the mast’ on the matter of niches: we must think of cephalosporins in these terms. It is possible to be pushed too far along these generations of cephalosporins; for example, to use third-generation cephalosporins when first- or second-generation ones would suffice. Indeed, at St Thomas Hospital, we pulled back from using third-generation and started using second-generation compounds, such as cefuroxime, fully successfully. Therefore, I want to support the idea of niches rather than a progression where the last one on the list is necessarily the best choice. I also support the concept that there should be no sharp divisions in cephalosporins. However, the damage has already been done, because it was commercially advantageous to divide cephalosporins into generations. The concept of generation implies that one generation should die and be replaced by the next, which is a good marketing exercise. Although it is true that there are certain resemblances between several groups, I fully support the concept of dividing not by generations but by chemical structure, because this often relates to antibiotic activity. I agree with previous speakers that the generation problem will last a long time. Usage of cephalosporins in Italy may be of interest. Italy is unique in the world—not only in Europe. In Italy, community-acquired lower respiratory infections are almost always treated with a third-generation cephalosporin, in contrast to Sweden, for example, where penicillin is usually used for pneumonia, or the UK, where amoxycillin is the antibiotic of choice. We have a low incidence of penicillin-resistant pneumococci, and the only similarity to other countries in Europe is the percentage of MRSA. The prevalence of MRSA is about the same level as in France. Is the use of third-generation cephalosporins in community-acquired infection perhaps responsible for the low incidence of resistance in Italy? First, regarding Professor Acar's comment on cephalosporins, it is true that the daily dose is not very high, but if you have an oral cephalosporin that is not a ‘pro-drug’, absorption is due to di-peptide or tri-peptide transporters. Thus, increasing the dose will decrease the bioavailability. For a pro-drug such as cefuroxime axetil, increasing the dose increases the side effects. It is a problem of tolerance. Second, concerning differences between third- and fourth-generation cephalosporins. A drug active against Staph. aureus will be of decreased activity against Pseudomonas. It is always a balance between both: very difficult to overcome. The best example is cefpirome, which is active against Pseudomonas (but less so than ceftazidime) and against Staph. aureus, similarly to cephalothin or cephaloridine. Macrolides are also good against Gram-positive cocci, but not against H. influenzae. It is difficult to target for both. Any comments on my remarks about cephalosporins vs. aminoglycosides? Can we now put away the aminoglycosides, because they are more toxic than the cephalosporins? We most often use aminoglycosides in combination—frequently with a cephalosporin—for either pseudomonal infections, for endocarditis and sometimes for Gram-negative sepsis in an immunocompromised individual. We have had very good experience with treatments lasting 7–10 days. Toxicity has not been an issue and, at least in North America and Canada, resistance rates to gentamicin in Gram-negative bacteria are still less than 1–2%, except with Pseudomonas-aeruginoa. Gentamicin is still standard therapy for endocarditis and, where third-generation cephalosporins are used as first-line therapy, as with the HACEK group of organisms, we will use ceftriaxone or cefotaxime. We increasingly use cephalosporins for endocarditis, such as for viridans streptococci, especially in out-patients who receive ceftriaxone once a day. I agree with Professor Baquero that most people believe that the latest generation is better than the previous one. A feature of cephalosporins that has not been addressed is their bactericidal activity. There are data that suggest that third-generation cephalosporins select for penicillin resistance in the pneumococcus. My question to the roundtable is: do you think that rates of bactericidal activity are something that we look at in terms of choosing our cephalosporins, because a dead bacterium does not become resistant? Bactericidal activity in third- or fourth-generation cephalosporins depends on the substituent on position 3, and decreases as you increase the size of the inoculum. That is well known for ceftazidime and applies to cefpirome, cefepime and chemically related compounds. Concerning the pneumococcus, I do not know if the use of cephalosporins increases the resistance rate but, again, bactericidal activity depends on the structure in position 3. Everybody says they want a narrow-spectrum antibiotic, but they all use broad-spectrum ones! We had this experience in France with cefsulodin, which has virtually only antipseudomonas activity: once ceftazidime was introduced to clinical practice, the sales of cefsulodin went down in 2 months, which is proof that people prefer a broad spectrum. Addressing Professor Baquero's remark: all the cephalosporins are not the same and, if you test only for cefotaxime, ceftriaxone or ceftazidime, you will never get the full picture. A good example is S. pneumoniae: when this is susceptible to penicillin, you have no problem but, when it is resistant to penicillin, only two or three cephalosporins are active. Therefore, you always have to be proactive and to test as many as possible. This kind of philosophy condemns clinical laboratories to anarchy. A diagnostic laboratory cannot test everything. We have got to undertake some rationalization. I think that, if you define bactericidal activity and all the rest of it in a research mode, then you have some guidance for the ordinary diagnostic laboratory. We tested a combination of ceftriaxone, cefotaxime and ceftazidime. The effect of the latter on PBP2 increased the bactericidal activity of all three compounds; the difference was quite striking. In France and Spain, injectable cephalosporins were confined to hospital use only for more than 15 years. Then ceftriaxone went outside hospitals, because it could be given once a day. It was the only cephalosporin outside the hospital. However, despite hospitals in France having the same cephalosporin consumption as in Spain, ESBLs appeared in France first in 1976–77, i.e. some 8 years earlier than in any other country. In Spain, there are still very few ESBLs; maybe they use more of one compound than another. Maybe we should not measure consumption in terms of cephalosporins in general; maybe we should use combinations. I think it was just by chance that one or several clones of ESBLs became common in France and disseminated among a number of hospitals. Another possibility is that the very high consumption of other types of antibiotic may keep away those plasmids coding for ESBLs. I agree that it is not just cephalosporin use but which cephalosporin; there is a difference in their ability to select for resistance. A second factor concerns whether they are used alone or in combination; and a third aspect is that, in Paris, there is a lot of exchange of patients from one hospital to another to receive different treatments and different diagnostic tests. I believe this does not happen in Spain. It is in this way that an ESBL can go from one hospital to another. I make the argument in the USA that we should try to maintain a certain independence and autonomy within units, because of the spread of resistant strains. Coming back to the classification of cephalosporins, while I agree with others that the system is flawed, for general practitioners (GPs) and for doctors working in hospitals who are not specialists in infectious disease, it can be quite useful. The so-called fourth generation of cephalosporins are not really the fourth generation but, in my personal opinion, are improved third-generation cephalosporins. Factors that affect the spread of resistance are far from clear. In Italy, there is considerable use of third-generation cephalosporins but, even in intensive care units (ICUs), we have very few ESBLs. In Bolivia in 1986, we found a considerable level of resistance related to the use and free sale of antibiotics by pharmacies, as did Dr Levy in Chile. This represents a paradox, in that it is such a poor country, but the only rich people are pharmacists, because drugs are sold so readily. We have heard a great deal about cephalosporin resistance in species that used to be susceptible–particularly about the emergence of ESBLs. However, another problem is the selection of those species that are inherently resistant to cephalosporins. Has our great usage of cephalosporins been a major factor behind the rise of the enterococcus and the Clostridium difficile? Would more use of beta-lactam/inhibitor combinations or carbapenems have made these problems less serious? Ten years ago, we started to notice that something was happening with enterococci; for example, they were found more frequently in urinary tract infections that originated in the community. Was it the use of cephalosporins, was it the introduction of quinolones or had the enterococcus suddenly acquired the ability to become epidemic? There gave been numerous searches for spurious correlations, which is very dangerous. However, there does seem to be a certain degree of logic to this linkage. We found that most of our hospital enterococci were ampicillin resistant and, of course, were Enterococcus faecium, and there were some outbreaks in our ICU. My colleague John Boyce carried out a case control study: the best correlation was with the use of carbapenems, and not the cephalosporins. In France, overall 14% of Klebsiella had ESBLs; 50% in Claremont Ferand; and 0% in a hospital in Normandy. Antibiotic resistance is local and, until we understand what the relationship is between local antibiotic usage and local emergence of resistance, we are going to be like the blind man feeling the tail of the elephant and trying to describe what the whole creature is like. Presumably, if we went back to the 1960s, there was a period when ampicillin resistance was local, but 40–60% of Escherischia coli now have beta-lactamase. Thus, over time, some averaging out does occur. Resistance to third-generation cephalosporins is still in the early phases, so is randomly located. It is a progression. A study at the National Institutes of Health (NIH) showed that ceftazidime was a perfectly good drug to treat leukemia in children. It should be remembered that the NIH is a very special place, where there is a limited number of patients, careful infection control, etc. However, when the Children's Hospital at Stanford University stopped using aminoglycosides and just used ceftazidime, they had an outbreak of TEM-26, with a few cases of TEM-12. As resistance developed, they recognized it, stopped the use of ceftazidime, went back to the old ways of using a beta-lactam plus aminoglycoside, and the problem disappeared. A couple of years later, in Leeds, England, they adopted the same philosophy, using ceftazidime: as before, there was an outbreak of resistance caused by TEM-12 and TEM-26. They sequenced the isolate and there were mutations away from the active site region, which designated that this was an example of convergent evolution. How the clone spreads depends on the local arrangement of medical services in the local community, but it first has to be selected by antibiotics, and we need to know more about that. Returning to the issue of C. difficile, many hospitals in the UK are finding this problem, so much so that some are introducing policies to reduce the use of cephalosporins, replacing them with either aminoglycosides or quinolones. My question to the panel is: which cephalosporin has the least ability to select C. difficile? Probably those that are essentially excreted in the urine should have less effect, but most of them do select for C. difficile. Local epidemiology can help us with the therapy but is completely useless in understanding the spread of resistance. Why is it that one ICU has a lot of beta-lactamase-producing strains and another ICU in the same town, with the same use of antibiotics has none? I spent some of my 17 years in Naples in a horrible old hospital—an old convent in terrible condition—and there was no hospital infection at all. After 7 years, we moved to a new hospital and immediately had an outbreak of a staphylococcal infection. This is not easy to understand. A recent publication in the Journal of Hospital Infection gave the odds ratios for various antibiotics causing C. difficile superinfection. Cefaclor is the least likely to cause the problem, because of its fragility in the gut. Cefaclor normally passes into the urine and the rest of it that gets into the gut is destroyed by Bacteroides beta-lactamases. Some multiresistant clones have a much greater ability to spread than others. Klebiella serotype K25 with SHV-4 enzyme (an ESBL) has run across a whole series of hospitals in France and Belgium. We do not know why it does better than a lot of the sporadic strains one sees in exactly the same centers. We do not know why a certain gene is successful and why a specific strain is successful. Perhaps they are better adapted to a particular niche, or they have something adapted to the niche that gives the best chance of jumping from one niche to the next. We could perhaps eventually be able to intervene in that process. The epidemigenicity of a given bacteria is proportional to its spread in host-to-host transmission. The best spreaders are those that have provoked infection. The best method of control is to detect the resistance problem as soon as possible and try to prevent the spread of infection. We have had a good discussion about the cephalosporins in the 1990s and many of the problems that afflict our patients. One thing is certain in the future, as Stuart Levy has told us, bacterial resistance will not go away. In the 1970s, I remember many people referring to the cephalosporins as ‘antibiotics in search of an indication’. In those days, of course, resistance was not the problem that it is today. There is little doubt that the cephalosporins have found their place in the treatment of many infections. It is uncertain whether we will see completely new cephalosporins attacking novel targets, but there is little doubt that they will continue to play a very important part in the management of serious infections, both inside and outside the hospital.