A Question of Balance

	INTRODUCTION

TOP INTRODUCTION MODIFYING THE HOST RESPONSE... PROINFLAMMATORY STRATEGIES TO... EFFECTS OF G-CSF IN... CLINICAL TRIALS WITH G-CSF... CONCLUSION REFERENCES

In this issue of the Journal, Karzai and colleagues (1) present the results of their studies in an animal model of pneumonia, suggesting that the type of bacterial infection and associated mediators released may be critical factors that determine the effectiveness of therapy with granulocyte colony-stimulating factor (G-CSF), a hematopoietic growth factor that increases both neutrophil number and function. Furthermore, the authors speculate that their observations may explain the results of recently published clinical trials with G-CSF in nonneutropenic patients with pneumonia, which to date have not shown "convincing benefit." It is my opinion that the data support another point of view.

	MODIFYING THE HOST RESPONSE TO INFECTION

TOP INTRODUCTION MODIFYING THE HOST RESPONSE... PROINFLAMMATORY STRATEGIES TO... EFFECTS OF G-CSF IN... CLINICAL TRIALS WITH G-CSF... CONCLUSION REFERENCES

Current therapy for serious infections includes antibiotic agents directed against the microorganism and supportive measures aimed at stabilizing and compensating for failing organs. The primary focus of the majority of new adjuvant therapies for serious infections such as sepsis has been the downregulation of the host inflammatory response to the precipitating insult (2). The hypothesis has been that the immune system of the infected patient is dysregulated and in order to preserve organ function and improve survival, the host response to the invader and/or its products must be attenuated. A large number of clinical trials, now spanning more than 15 yr of investigation, which have attempted to suppress the inflammatory response have all failed to show benefit in this critically ill patient population.

Why have these trials failed to show clinical benefit? Several explanations have been offered, including the timing of the intervention, the selection of the appropriate patient population, the dose and duration of the therapy, and the fact that the preclinical data that provided the rationale for the clinical trials were both limited and, perhaps, not relevant to the clinical situation (3). Furthermore, the basic hypothesis that "rogue inflammation" was the appropriate target in these infected patients may, in fact, have been wrong (4). Relevant to this issue is the observation that the only therapeutic modality to date that has shown any benefit in improving patient survival in patients with documented bacteremia is antibiotic therapy specifically directed to the offending pathogen (5). Why is it, therefore, logical to assume that attempts to downregulate the response of the host to the offending pathogen would result in clinical benefit?

	PROINFLAMMATORY STRATEGIES TO ENHANCE HOST DEFENSE

TOP INTRODUCTION MODIFYING THE HOST RESPONSE... PROINFLAMMATORY STRATEGIES TO... EFFECTS OF G-CSF IN... CLINICAL TRIALS WITH G-CSF... CONCLUSION REFERENCES

One could reasonably argue that if unresolved infection is the precipitating factor that drives the systemic inflammatory response resulting in septic shock and multisystem organ failure, then amplification of the host defense system, rather than its attenuation, may more successfully resolve the inciting event i.e., the infectionand prevent its sequelae. Furthermore, perhaps the host defense system of the septic patient, much like the cardiovascular and other organ systems, may represent yet another failing organ system. Under these circumstances, an antiinflammatory strategy may only serve to compromise further the remaining vestiges of the patient's immune system. On the other hand, there likely is a point beyond which attempts to stimulate and salvage the immune system are no longer effective.

One promising strategy for upregulating the host defense system of the infected patient focuses on the use of G-CSF. Colony-stimulating factors are glycoproteins that are required for the proliferation and differentiation of hematopoietic progenitor cells. Of this cytokine family, it is G-CSF that plays a critical role in determining both the number and biologic activities of neutrophils (6). The ability to selectively modulate both the number and functional state of neutrophils renders G-CSF an attractive candidate for immunotherapy in patients with serious infections. The fundamental question is whether exogenous G-CSF can promote the recovery of the nonneutropenic patient from either a local or sytemic infection by enhancing the activity of pre-existing neutrophils and/or by increasing the number of these effector cells.

	EFFECTS OF G-CSF IN NONNEUTROPENIC MODELS OF INFECTION

TOP INTRODUCTION MODIFYING THE HOST RESPONSE... PROINFLAMMATORY STRATEGIES TO... EFFECTS OF G-CSF IN... CLINICAL TRIALS WITH G-CSF... CONCLUSION REFERENCES

Numerous animal studies have investigated the effects of G-CSF in a variety of infection models in animals. In the majority of these studies, G-CSF has been shown to increase the production of neutrophils in a dose-dependent fashion, enhance the delivery of neutrophils to the site of infection, reduce the burden of infection, and improve survival either alone or in combination with antibiotic therapy (6). However, as in the study in this issue of the Journal, these beneficial effects in preclinical models of infection have not been uniformly observed. One such study, by Terashima and coworkers (7), is particularly relevant to the present investigation by Karzai and colleagues. In this study the investigators rendered one group of guinea pigs granulocytopenic by pretreating them with cyclophosphamide and rendered another group neutrophilic by giving them G-CSF. In a third, control group, the neutrophil count was not manipulated. All of the animals were then given either a low dose (10⁴ colony-forming units [CFU]/ kg) or a high dose (10⁸ CFU/kg) of Pseudomonas aeruginosa via intratracheal instillation. The investigators found that animals with G-CSF-induced neutrophilia did better than the other two groups of animals when given the low-dose inoculum of P. aeruginosa. In contrast, 70% of G-CSF-treated animals died within 4 h when challenged with the high-dose inoculum. The authors concluded that neutrophils protect against lung injury during low-level bacterial challenge but increase lung injury and contribute to mortality during a high-inoculum bacterial challenge.

However, another interpretation of this study is possible: that the "high-dose" intratracheal challenge used in these experiments is more representative of a model of lung injury than of bacterial pneumonia. In the normal animals, intratracheal instillation of the low dose (10⁴ CFU) of P. aeruginosa caused neutrophils to migrate into the lung, as would be expected as this is a critical part of the normal lung host defense response. In the animals with cyclophosphamide-induced granulocytopenia, fewer neutrophils were recruited into the lung. Many more neutrophils entered the alveoli of G-CSF-treated animals in response to the P. aeruginosa lung challenge almost a 15-fold difference, compared with the cyclophosphamide-treated animals. With the high-dose inoculum (10⁸ CFU) of P. aeruginosa, the number of neutrophils in the control animals increased, and it increased still further in animals that were pretreated with G-CSF. Granulocytopenic animals still had the fewest neutrophils in their lungs, although the number also increased in response to the high-dose inoculum.

The investigators also determined the recovery of live bacteria from the lungs of each group of animals. Among the animals that received the low-dose inoculum of P. aeruginosa, 2 of 11 control animals and 6 of 9 animals that received cyclophosphamide had lung cultures positive for Pseudomonas. The lungs of all of the G-CSF-treated animals were sterile. Thus, with a low-dose inoculum, G-CSF enhanced neutrophil recruitment into the lung, which accelerated bacterial clearance.

With the high-dose inoculum, none of the infections cleared in any of the study groups. Those given the high-dose inoculum and G-CSF died more rapidly than the controls or the cyclophosamide-treated animalswithin 4 h of the challenge. I would suggest that these animals died not from infection, but from lung injury. Similarly, in the study by Karzai and colleagues, I believe that the authors were, in fact, studying a "low-dose" versus a "high-dose" inoculum as well as the capacity of these two bacterial species to elicit an intrapulmonary inflammatory response. The fact that the authors used a dose of Escherichia coli that was approximately fivefold higher compared with the Staphylococcus aureus inoculum substantially argues against their contention that their experimental observations were "bacteria specific." In point of fact, the immune responses tested in this model system are by definition "nonspecific." Survival benefits with G-CSF treatment have been shown in several preclinical models of pneumonia with a broad spectrum of gram-negative pathogens, including P. aeruginosa, Pasteurella multocida, and Klebsiella pneumoniae (8). It is almost certain that the cause of the increased mortality that the authors observed with the E. coli challenge was not dependent on a live, proliferating bacterial infection but rather on the intrinsic virulence factors associated with that challenge and the resultant inflammatory response in a normal host whose neutrophil response was further amplified by pretreatment with G-CSF. This then resulted in an injured lung, which allowed for bacteria and/or their products along with locally produced mediators to enter the systemic compartment and promote a systemic inflammatory response. In support of the concept that the model studied in this article is more representative of lung injury than infection, the authors report elevated levels of tumor necrosis factor  (TNF) in the serum of the E. coli-challenged rats. Other investigators, including those of our laboratory, have reported in both animal models and in patients that during pneumonia TNF is compartmentalized within the lung and is not detected in the systemic circulation in the absence of lung injury (9).

Clearly, the inflammatory cascade operates in a delicate balance, and augmenting the number and/or functions of neutrophils may exacerbate organ injury under certain conditions. We examined this possibility in two noninfectious experimental models of lung injury in rats (12). In one model, acute lung injury was induced by systemic administration of -naphthylthiourea, an agent that causes neutrophil-independent oxidant injury to endothelial cells. In another model, intrapulmonary hydrochloric acid caused a neutrophil-dependent injury. Not surprisingly, the results showed that augmentation of neutrophil number or function (or both) by pretreatment with G-CSF in these models of sterile inflammation potentiated lung damage. Thus the role of neutrophils in either host defense or tissue injury is determined by the environment into which they migrate, and neutrophil responses must be understood in the context of specific conditions and specific tissues.

	CLINICAL TRIALS WITH G-CSF IN NONNEUTROPENIC PATIENTS WITH PNEUMONIA

TOP INTRODUCTION MODIFYING THE HOST RESPONSE... PROINFLAMMATORY STRATEGIES TO... EFFECTS OF G-CSF IN... CLINICAL TRIALS WITH G-CSF... CONCLUSION REFERENCES

Increasing the neutrophil count in a normal animal and then introducing an overwhelming proinflammatory insult into the lung may well result in neutrophil-mediated lung injury. It is doubtful, however, that the same process would likely occur in infected patients who might be candidates for treatment with G-CSF. In fact, no such observation has been made in the approximately 1,000 nonneutropenic pneumonia patients treated with G-CSF in randomized, placebo-controlled clinical trials. While I agree that published trials with G-CSF have not shown "convincing benefit" as of yet, I believe that this is not due to pathogen-specific effects as proposed by Karzai and co-workers but rather is due to the severity of illness of the patients studied as well as other factors.

Clearly one of the primary concerns in administering G-CSF to nonneutropenic patients with pneumonia was patient safety. Conventional wisdom predicted that these patients would likely develop adult respiratory distress syndrome (ARDS) and possibly other organ failures. Therefore, the first large clinical trial of filgrastim (r-metHuG-CSF; Amgen, Thousand Oaks, CA) in pneumonia was designed to include hospitalized patients with mild to moderate disease. This was a multicenter, double-blind, randomized, parallel-group, placebo-controlled study of 756 hospitalized patients with community-acquired pneumonia (13). Overall, length of stay (7 d) and mortality (6%) were not affected by filgrastim therapy. Notably, filgrastim-induced neutrophilia was not associated with worsening of blood oxygenation or chest radiograph, underlying safety concerns that were important in the design of this trial. On the contrary, analysis of safety data showed that filgrastim treatment reduced the occurrence of systemic complications of pneumonia. When analyzed on an intent-to-treat basis, the incidences of ARDS and disseminated intravascular coagulation occurred significantly less frequently with filgrastim treatment. Furthermore, analysis of data for a large subset of patients (261 patients) with more severe illness, that is, those prospectively identified to have multilobar pneumonia, also showed a reduction in the incidence of both local and systemic complications. Finally, there were no differences in mortality or adverse events in these pneumonia patients in terms of the causative organism. This study is significant, as it represents the first large trial of filgrastim in the treatment of infection in nonneutropenic patients. In contrast to other trials that have focused on suppressing the inflammatory response in infected patients, this trial focused on enhancing host defenses.

Whether the potential benefits of filgrastim therapy are due solely to an increase in the number of neutrophils or also result from enhancing the function of these effector cells remains unclear at this time. Furthermore, it is simplistic to assume that all patients with a variety of comorbidities and pathogens presenting at different stages of their disease will respond to G-CSF therapy in a uniform fashion. We need to further our understanding of how underlying illnesses and their attendant therapies disrupt normal host defense mechanisms if we are to develop effective cytokine-based therapies. Also, the choice of antibiotic therapy may be of importance when using cytokines as adjuvant therapy. The therapeutic efficacy of certain antibiotics is affected by the number of blood neutrophils present at the time of infection (14) and G-CSF has also been reported to increase the uptake of certain antibiotics by neutrophils (15). As G-CSF has been shown in animal models to increase the number of neutrophils that are delivered to an infected tissue site, one could potentially optimize therapy by "targeting" antibiotic delivery to an infected site by choosing an appropriate antibiotic in combination with G-CSF. This would be particularly useful in parts of the body where adequate antibiotic concentrations might be difficult to achieve and/or if the patient is infected with a relatively resistant pathogen. In addition, the primary focus of recent trials with G-CSF has been on the treatment of infected patients, while the majority of preclinical studies showing benefit have administered G-CSF before the infectious challenge. Finally, the development of a multimodal approach, including components of immune modulation and immune restoration, is likely needed to correct the multiple aberrations in the host defense system that occur in our critically ill patients.

	CONCLUSION

TOP INTRODUCTION MODIFYING THE HOST RESPONSE... PROINFLAMMATORY STRATEGIES TO... EFFECTS OF G-CSF IN... CLINICAL TRIALS WITH G-CSF... CONCLUSION REFERENCES

The host defense system operates in a delicate balance that functions to maintain homeostasis during an infectious episode. Because of their role in mediating critical aspects of immunity, cytokines, such as G-CSF, have a great potential for reducing the morbidity and mortality caused by infections and have begun to find their way into clinical trials. However, it is best to proceed with caution, as efforts to stimulate the immune system nonselectively may prove to be as deleterious to the patient as the negative effects of their immunocompromised state. As the complexities of the host-pathogen interaction are further dissected and unraveled, it is likely that the therapeutic benefits from these immunomodulators will be fully realized.

STEVE NELSON

John H. Seabury Professor of Medicine

Pulmonary/Critical Care Medicine

Louisiana State University Medical Center

New Orleans, Louisiana

	References

TOP INTRODUCTION MODIFYING THE HOST RESPONSE... PROINFLAMMATORY STRATEGIES TO... EFFECTS OF G-CSF IN... CLINICAL TRIALS WITH G-CSF... CONCLUSION REFERENCES

1. Karzai, W., B. U. von Specht, C. Parent, J. Haberstroh, K. Wollersen, C. Natanson, S. M. Banks, and P. Q. Eichacker. 1999. G-CSF during E. coli versus S. aureus pneumonia in rats has fundamentally different and opposite effects. Am. J. Respir. Crit. Care Med. 159: 1377-1382 [Abstract/Free Full Text].

2. Natanson, C., W. D. Hoffman, A. F. Suffredini, P. Q. Eichacker, and R. L. Danner. 1994. Selected treatment strategies for septic shock based on proposed mechanisms of pathogenesis. Ann. Intern. Med. 120: 771-783 [Abstract/Free Full Text].

3. 1997. From the bench to the bedside: the future of sepsis research. Executive summary of an American College of Chest Physicians, National Institute of Allergy and Infectious Disease, and National Heart, Lung, and Blood Institute Workshop. Chest 111:744-753.

4. Bone, R. C.. 1996. Commentary: why sepsis trials fail. J.A.M.A. 276: 565-566 [Abstract/Free Full Text].

5. Bryan, C. S., K. L. Reynolds, and E. R. Brenner. 1983. Analysis of 1,186 episodes of gram-negative bacteremia in non-university hospitals: the effects of antimicrobial therapy. Rev. Infect. Dis. 5: 629-638 [Medline].

6. Dale, C. D., W. C. Liles, W. R. Summer, and S. Nelson. 1995. Review: granulocyte colony-stimulating factorrole and relationships in infectious diseases. J. Infect. Dis. 172: 1061-1075 [Medline].

7. Terashima, T., M. Kanazawa, K. Sayama, T. Urano, F. Sakamaki, H. Nakamura, Y. Waki, K. Soejima, S. Tasaka, and A. Ishizaka. 1995. Neutrophil-induced lung protection and injury are dependent on the amount of Pseudomonas aeruginosa administered via airways in guinea pigs. Am. J. Respir. Crit. Care Med. 152: 2150-2156 [Abstract].

8. Nelson, S., and W. R. Summer. 1998. Innate immunity, cytokines, and pulmonary host defense. Infect. Dis. Clin. N. Am. 12: 555-567 [Medline].

9. Nelson, S., G. J. Bagby, B. G. Bainton, L. A. Wilson, J. J. Thompson, and W. R. Summer. 1989. Compartmentalization of intra-alveolar and systemic lipopolysaccharide-induced tumor necrosis factor and the pulmonary inflammatory response. J. Infect. Dis. 159: 189-194 [Medline].

10. Dehoux, M. S., A. Boutten, J. Ostinelli, N. Seta, M. C. Dombret, B. Crestani, M. Deschenes, J. L. Trouillet, and M. Aubier. 1994. Compartmentalized cytokine production within the human lung in unilateral pneumonia. Am. J. Respir. Crit. Care Med. 150: 710-716 [Abstract].

11. Tutor, J. D., C. M. Mason, E. Dobard, R. C. Beckerman, W. R. Summer, and S. Nelson. 1994. Loss of compartmentalization of alveolar tumor necrosis factor after lung injury. Am. J. Respir. Crit. Care Med. 149: 1107-1111 [Abstract].

12. King, J., B. P. deBoisblanc, C. M. Mason, J. M. Onofrio, G. Lipscomb, D. E. Mercante, W. R. Summer, and S. Nelson. 1995. Effect of granulocyte colony-stimulating factor on acute lung injury in the rat. Am. J. Respir. Crit. Care Med. 151: 302-309 [Abstract].

13. Nelson, S., S. M. Belknap, R. W. Carlson, D. Dale, B. deBoisblanc, S. Farkas, N. Fotheringham, H. Ho, T. Marrie, H. Movahhed, R. Root, J. Wilson, and for the CAP Study Group. 1998. A randomized controlled trial of filgrastim as an adjunct to antibiotics for treatment of hospitalized patients with community-acquired pneumonia. J. Infect. Dis. 178: 1075-1080 [Medline].

14. Yasuda, H., Y. Ajiki, T. Shomozato, M. Kasahara, H. Kawada, M. Iwata, and K. Shimizu. 1990. Therapeutic efficacy of granulocyte colony-stimulating factor alone and in combination with antibiotics against Pseudomonas aeruginosa infections in mice. Infect. Immun. 58: 2502-2509 [Abstract/Free Full Text].

15. McKenna, P. H., S. Nelson, and J. Andresen. 1996. Filgrastim (rhuG-CSF) enhances ciprofloxacin uptake and bactericidal activity of human neutrophils in vitro (abstract). Am. J. Respir. Crit. Care Med. 153S: A535 .