General Principles

Cytotoxic therapy–induced neutropenia is a major predisposition to infection.⁶ Neutrophil counts below 1000 cells/µL (the total absolute number of polymorphonuclear neutrophils plus bands) increase susceptibility to infection in a linear fashion (i.e., the lower the neutrophil count, the greater the degree of susceptibility)¹⁸ (see Fig. 53.1). Although most research studies use 500 cells/µL as an arbitrary definition of neutropenia, intensivists must recognize that susceptibility increases as the neutrophil count declines below 500 to 1000 cells/µL. A patient with a neutrophil count of 100 cells/µL is much more vulnerable to infection than a patient with 500 or 1000 cells/µL, and a patient with zero neutrophils is at much higher risk for fulminant infection than a patient with 50 or 100 cells/µL. The trajectory of the neutrophil count is also important: a patient with a neutrophil count of 1500 cells/µL whose counts are dropping precipitously should best be treated like a patient with absolute neutropenia. Similarly, a patient with 500 neutrophils/µL whose counts are rising quickly is not nearly as vulnerable to a poor outcome as a patient with a count of 500 neutrophils/µL that is stable.

Patients with neutropenia are generally divided into high-risk and low-risk patients based on their likelihood of developing severe infectious complications. Markers for high risk include neutropenia for more than 7 days’ duration and neutrophil count less than 100 cells/µL, as well as obvious signs of a life-threatening process such as hypotension, obtundation, pneumonia, or severe abdominal pain. As Figures 53.2 to 53.4 outline, this risk assessment is used in designating empiric regimens.

Thus, although the absolute neutrophil count is an essential factor to follow, the duration of neutropenia, the functional capability of neutrophils, the integrity of physical barriers such as the skin and gastrointestinal mucosa, the patient’s microbiologic environment (endogenous and exogenous flora), and the status of other immune mechanisms also contribute to the infectious syndromes that will develop.

In the 1960s and 1970s, aerobic gram-negative bacilli such as Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa predominated as pathogens in neutropenic patients. In the 1990s the spectrum of causative pathogens in neutropenic patients shifted from a predominance of gram-negative bacilli to a majority of gram-positive cocci including streptococci, staphylococci (including oxacillin-resistant Staphylococcus aureus), and enterococci (including vancomycin-resistant enterococci).10,12–15,19–21 Candida species have also become more frequent as pathogens, especially as patients are on broad-spectrum antibacterials and have long-term venous access devices in place.

More recently, highly resistant gram-negative bacilli have become major threats for nosocomial transmission. Clinicians must consider the possibility that a patient may be colonized and then infected with a Stenotrophomonas, a Burkholderia, or a carbapenemase-producing gram-negative Enterobacteriaceae such as a Klebsiella, an Enterobacter, or an E. coli that has developed mechanisms that evade currently marketed drugs.22–26

The management of febrile, neutropenic fever is reviewed in a guideline that is widely used to direct care in North America.⁶ Figures 53.2 to 53.4 summarize important aspects of management. Table 53.3 also provides a summary of useful management information. Table 53.4 outlines common prevention strategies that will modify the spectrum of causative pathogens. Box 53.1 summarizes the organisms that most often cause disease in neutropenic patients.

Patients with neutropenia are generally suspected of being infected if the clinical syndrome is consistent with infection, or if the temperature is at least 38.3° C on one occasion or 38.1° C on two separate occasions. An elevated temperature alone should trigger the institution of broad-spectrum antimicrobials in almost all situations. Given this emphasis of using temperature as an indicator for starting antimicrobial therapy, using a validated technique to measure temperature is important. Although pulmonary artery or urinary catheter thermistors appear to provide the most accurate measurement, most experienced ICUs use tympanic membrane thermistors. Rectal probes are avoided in order to reduce the induction of perirectal infections in neutropenic patients, and to reduce the potential for fecal pathogen transmission. As noted earlier, although fever is almost always a reason to start antimicrobial therapy in this patient population, the absence of fever should not be the grounds for avoiding antimicrobials if a patient has other symptoms or signs suggesting infection. The threshold for starting antimicrobials should be very low, i.e., if there is a suspicion of infection, a broad-spectrum regimen should be started.

As noted earlier, the initial regimen should not be parsimonious in terms of spectrum. Because this population of patients is susceptible to a wide variety of bacterial and fungal pathogens, a very wide broad-spectrum regimen should be used. There are many potential regimens, each of which must be tailored to the local experience with the patient population and the hospital, specific patient factors such as evidence of prior colonization or recent antimicrobial therapy, and clinical manifestations suggesting infection. Popular regimens would include (1) meropenem or cefepime or piperacillin-tazobactam for broad-spectrum antibacterial activity plus (2) vancomycin or daptomycin for staphylococcal infections27,28 plus (3) ciprofloxacin or moxifloxacin or aztreonam for broader gram-negative bacillus coverage. Many experienced clinicians would add an echinocandin for anti-Candida activity given the frequency of intravascular catheter-associated infections due to Candida species.^29–33 Intensivists need to work closely with their infectious disease consultants, microbiology laboratories, and referring teams to develop regimens that are optimal for their hospital environment, for the patient population involved, and for the specific, unique patient who is being managed.

Empiric antiviral therapy is not usually initiated unless there is a specific reason to suspect a viral process. Antiviral agents would generally be added only if a specific viral process such as CMV colitis or disseminated herpes simplex were suspected.

For the duration of neutropenia (when the neutropenia is expected to be time-limited), broad-spectrum therapy must be continued. When a specific causative organism is identified, antimicrobial therapy should be optimized for that organism. However, unlike other immunologically normal patients, the broad-spectrum “background” (i.e., meropenem or piperacillin-tazobactam or cefepime or ceftazidime) must be continued until the neutropenia resolves, on the assumption that a patient who develops one infection is likely to develop or manifest another infectious process while the neutropenia persists.

Coverage for methicillin-resistant Staphylococcus aureus (MRSA) does not necessarily need to be continued if no MRSA is identified. There are environmental advantages to reducing vancomycin exposure.

As noted previously, there is no documented reason, even in this population, to use combination therapy to treat a specific pathogen in most situations, although combination therapy is needed for the empiric approach for the duration of neutropenia (Box 53.2). In rare situations, if the causative organism is not susceptible to agents with well-documented efficacy, combination therapy may be an appropriate strategy out of desperation. As an example, for treating enteric carbapenemase-producing organisms, a combination of tigecycline plus colistin plus an aminoglycoside might be desirable given the high minimum inhibitory concentrations for all antibiotics for these organisms and the dreadful clinical results with any therapeutic intervention.^34–38

For patients with fever and neutropenia, the cause of the fever is historically documented in only 50% to 60% of patients. The duration of therapy once empiric antimicrobials are started depends on the evolution of the neutrophil count, the patient’s clinical status, and the results of diagnostic tests. If the patient defervesces and looks clinically well, however, the broad-spectrum regimen should not be stopped until the neutrophil count is above 500 to 750 cells/µL, preferably on two occasions, and the patient has received at least 10 to 14 days of therapy.

If a neutropenic patient is started on antibacterial therapy without fungal therapy, and defervescence has not occurred by days 3 to 5, an antifungal agent should be added (Figs. 53.3 and 53.4). The choice of antifungal agent depends on the patient population and the patient’s specific history. In the current era many patients have been receiving short- or long-term prophylaxis with fluconazole, voriconazole, or posaconazole. Although in general clinicians could add fluconazole, an echinocandin (e.g., caspofungin or micafungin or anidulofungin) or liposomal amphotericin B can also be used. An echinocandin is often a preferred choice if the patient has been on long-term azole prophylaxis and if a mold infection is not suspected.^39–42

As patients receive chemoprophylaxis with quinolones or azoles during periods of intense neutropenia or immunosuppression, breakthrough pathogens are more and more likely to be resistant to the prophylactic agents.40,42 Thus empiric regimens must be chosen with keen attention to the drugs that patients have received in the recent past, as well as pathogens they have previously been colonized or infected with.

Diagnostic Approach

Patients with fever and neutropenia require aggressive diagnostic efforts to identify the cause of fever so that the appropriate antimicrobial agent is used and appropriate procedures (e.g., surgical drainage, removal of a foreign body such as a catheter) can be performed. All febrile neutropenic patients should at a minimum have two blood cultures drawn, with one drawn peripherally and one drawn through the lumen of the indwelling catheter that has either been in the longest or is most suspicious for being infected. Other sites should be cultured as clinically indicated.

Regular physical examination is necessary to identify sites that merit more focused investigation. With impaired inflammatory response, findings on examination may be subtle. Knowledge of the specific immunologic defect is important so that when cultures of blood, sputum, urine, or other appropriate body fluids or body sites are performed, special microbiologic approaches can be used to detect viruses, fungi, helminths, protozoa, and bacteria as indicated by the clinical situation. Imaging studies are also important because intra-abdominal, intrathoracic, intracerebral, and musculoskeletal processes can be clinically subtle and may not be associated with identifiable organisms in the bloodstream. A growing array of antigen detection systems and molecular and high-performance chromatographic tests are being investigated to facilitate diagnosis.

Some of these approaches, despite their promising initial reports, are not yet clinically practical because of their level of sensitivity, specificity, or the cost or expertise required to perform them adequately. For instance, the PCR test for Pneumocystis is so sensitive that there is no clear separation of patients who are colonized with Pneumocystis (and whose pulmonary dysfunction is due to another process), and the serum β-glucan antigen detection system is so nonspecific that some clinicians are not confidant that the test provides useful information.43–45 Similarly, the PCR test for respiratory syncytial virus (RSV) or influenza or parainfluenza is so sensitive that immunosuppressed patients may shed small quantities of virus for many weeks after acute infection, confusing the diagnosis of the new pulmonary processes that occur after the acute viral infection is over, and at a time when another process is causing fever or pulmonary manifestations. Thus, these new tests must be interpreted with caution.

Empiric and Specific Antimicrobial Therapy

Outside the ICU, stable patients with fever and neutropenia, and no obvious source of infection, are treated with a broad-spectrum regimen, as mentioned previously, that covers all likely pathogens. Recommended regimens for the “backbone” agent include a carbapenem with antipseudomonas activity (e.g., meropenem or imipenem), a β-lactam β-lactamase combination with antipseudomonas activity (e.g., piperacillin-tazobactam), and a broad-spectrum cephalosporin with antipseudomonas activity (e.g., cefepime or ceftazidime). Although vancomycin is not necessarily indicated for empiric therapy of fever and neutropenia, in the ICU an antistaphylococcal drug (e.g., vancomycin or daptomycin) is usually added if the patient has a long-term intravascular catheter in place and may be appropriate empirically in every patient who merits ICU admission until the causative pathogen is known. Because patients with fever and neutropenia have almost always had extensive exposure to hospital-acquired pathogens and to antimicrobials, clinicians must adjust the empiric regimens to fit the patient’s situation. Adding colistin plus tigecycline empirically might be appropriate for someone hospitalized during an outbreak of highly resistant Acinetobacter or Klebsiella.

Patients in the ICU are by definition either unstable hemodynamically or medically fragile due to concurrent disease. In such situations, many clinicians would expand antibacterial coverage with a second broad-spectrum drug (e.g., a quinolone such as ciprofloxacin), aztreonam, or an aminoglycoside (e.g., gentamicin or tobramycin).20,46–51

A substantial number of febrile, neutropenic patients fail to improve in terms of fever or other manifestations. Once febrile and neutropenic patients are started on empiric therapy, if no causative process or organism is detected, one of the first three scenarios listed here is likely to be encountered. Figures 53.2 to 53.4 list some of the therapeutic options for such patients.

A common problem in febrile, neutropenic patients is managing indwelling intravascular lines.57,58 In general, these lines may be left in place initially if examination of the site reveals no indication of infection and the patient is hemodynamically stable. Blood cultures should be drawn through the catheter. Although some experts advocate drawing a culture through each port of each catheter, obtaining this many blood cultures is often not feasible because of time, cost, and volume of blood. If a patient is hemodynamically unstable and fails to respond promptly to fluid administration, it is prudent to remove the line in case an infected catheter is the source of the sepsis. Failure to remove the foreign body in this situation probably increases the likelihood of an unfavorable outcome. Should blood cultures become positive and should the suspicion be high that the catheter is the source, antibacterial therapy may be successful in some settings (e.g., if the pathogen is a bacterium that is relatively sensitive to antibacterial therapy), thus avoiding the need to remove the catheter. Situations suggesting that catheter removal is necessary include hemodynamic instability despite aggressive fluid resuscitation, tunnel infection, or infections resulting from fungi or relatively antibiotic-resistant bacteria such as P. aeruginosa.

Granulocyte transfusions have not been proved in randomized trials to improve survival in clinical settings probably because of the inability to administer a large number of cells with adequate frequency.16,17 However, many clinicians are convinced that matched white blood cell transfusions are helpful in managing life-threatening infections when patients are neutropenic and will use them when such cells are available. The manipulation of immune response with cytokines, cytokine inhibitors, or immunoglobulins is the subject of considerable investigation: Such interventions may reduce the duration of fever or the incidence of infections when used empirically, but in no setting have they been clearly shown to improve survival when administered after an infection has been documented. Algorithms for managing fever in neutropenic patients are provided in Figures 53.2 to 53.4. Table 53.3 suggests modifications of standard empiric regimens in certain common clinical scenarios.

Prevention of Infection

Given the experience with frequent and severe infectious complications in cancer patients with neutropenia, it has been logical to attempt to prevent infection. Most microorganisms causing disease in this patient population arise from endogenous gastrointestinal, cutaneous, or respiratory flora. Total protected environments probably reduce frequency of infection, but this approach is expensive and inconvenient. Trying to prove a consistent beneficial impact on survival has been difficult, and thus such isolation is rarely used anymore. Some experts are enthusiastic about placing patients in positive-pressure rooms so that pathogens do not enter via particles and droplets from outside the room. This type of isolation has not clearly improved outcome, however, and is not a standard of care.59,60 In Europe, there is more enthusiasm for such an approach than in the United States. Controversies over interpretation of data and concern that such antibiotic pressure will encourage the development of drug-resistant bacteria and fungi have diminished widespread acceptance in the United States.

Systemic antibacterial prophylaxis and systemic antifungal prophylaxis have been shown in some studies to reduce the number of infections, but their lack of effect on patient survival, their cost, and their impact on the emergence of resistance have made many clinicians reluctant to use them. Anti-Pneumocystis prophylaxis is, in contrast, highly effective in susceptible populations. Prophylaxis for CMV is rarely used unless the patient has received a solid organ or human stem cell transplant. Table 53.4 summarizes general strategies of infection prevention in immunosuppressed patients including patients with neutropenia.

Spectrum of Clinical Manifestations

Patients with HIV infection develop clinical disease as a result of three basic processes: (1) opportunistic infections and tumors that are enabled by HIV-induced immunosuppression; (2) the direct effect of HIV on specific organs (e.g., cardiomyopathy, enteropathy, dementia); or (3) immunologically mediated processes (e.g., glomerulonephritis, thrombocytopenia).

Management of Antiretroviral Drugs

For any HIV-infected patient in the ICU, clinicians must be cognizant of the need for careful management of antiretroviral drugs.⁷¹ If patients are not receiving antiretroviral drugs at the time of ICU admission, the ICU is not a desirable setting for initiating them: patient commitment to long-term adherence is difficult to assess when patients are critically ill, and drug toxicities and interactions will be hard to assess.^72,73 There is virtually no indication to start antiretroviral therapy acutely in the ICU following the diagnosis of an acute opportunistic infection with the rare exception of rapidly progressive forms of untreatable diseases such as JC virus encephalitis or cryptosporidiosis.

If patients enter the ICU already receiving antiretroviral therapy, an expert in HIV management should be consulted about the benefits of continuing the drugs rather than interrupting them. There is ample evidence that even brief interruption of antiretroviral drugs can have deleterious effects in terms of long-term loss of CD4 cells and in terms of the occurrence of opportunistic infections. However, administration of antiretroviral drugs in the ICU is challenging. Almost all commonly used antiretroviral drugs are available only in oral formulations, and thus absorption is often uncertain in critically ill patients. Subtherapeutic concentrations of antiretroviral drugs can select for drug resistance mutations, producing virus that is irrevocably nonsusceptible to the suboptimally dosed drug, sometimes with cross-class resistance. Many antiretroviral drugs affect cytochrome P-450 hepatic enzymes, resulting in altered pharmacokinetics for many non-AIDS-related drugs, which may substantially alter the efficacy or safety of the non-AIDS drug. Similarly, non-AIDS-related drugs can alter antiretroviral drug kinetics, resulting in drug serum levels that are above or below therapeutic targets, leading to viral resistance or drug toxicity. Stopping antiretroviral drugs may be the least harmful option. However, given the different half-lives of various antiretroviral agents, such discontinuation should be done in consultation with an HIV-experienced clinician.

Clinicians should refer to the Guidelines for Prevention and Treatment of Opportunistic Infections in HIV-Infected Adults and Adolescents and the Guidelines for Antiretrovial Therapy in Adults and Adolescents for more detailed discussion on when to initiate antiretroviral therapy in the setting of a specific opportunistic infection.⁷⁴

Diagnosis of Opportunistic Infections

The CD4⁺ T lymphocyte cell number continues to be a useful marker for predicting the occurrence of opportunistic infections in patients with HIV infection.⁵ This relationship of CD4⁺ T lymphocyte count to the occurrence of opportunistic infection continues to be as valid in the era of antiretroviral therapy as it was before the licensing of the first antiretroviral agent, zidovudine, in 1987.⁷⁵ Figure 53.5 demonstrates the typical relationship of CD4⁺ T lymphocyte counts to the occurrence of opportunistic infections. Knowledge of this relationship permits the focusing of diagnostic, therapeutic, and prophylactic management.

For instance, if a patient with HIV infection and a CD4⁺ T lymphocyte count of 700 cells/µL presents with diffuse pulmonary infiltrates, the diagnostic evaluation and empiric antimicrobial regimen should focus on common, nonopportunistic pathogens such as Mycoplasma, Legionella, and Chlamydia organisms, as well as common community-acquired viruses such as influenza plus opportunistic infections that occur at high CD4⁺ T lymphocyte counts, such as Mycobacterium tuberculosis or S. pneumoniae. In contrast, if the same patient had a CD4⁺ T lymphocyte count fewer than 50 cells/µL, the evaluation and empiric regimen would focus on pneumocystosis, pneumococcal pneumonia, and tuberculosis, although the previously mentioned processes that occur at high CD4⁺ T lymphocyte counts can also occur at lower CD4⁺ T lymphocyte counts.

Keeping in mind that CD4⁺ T lymphocyte counts are useful predictors of susceptibility to infection is important, but they are not perfect. Occasionally, patients will develop opportunistic infections at “uncharacteristically” high CD4⁺ T lymphocyte counts. For instance, 5% to 10% of cases of pneumocystosis occur at CD4⁺ T lymphocyte counts greater than 200 cells/µL.⁷⁶ Clinical parameters can provide additional clues; for example, oral candidiasis, a previous opportunistic infection, a prior episode of pneumonia, or high viral load are independent risk factors for the occurrence of Pneumocystis jiroveci pneumonia (PCP), and logically for other infections as well.

A frequent question is whether an HIV-infected patient’s prior CD4⁺ T lymphocyte count nadir affects the likelihood of an opportunistic infection occurring if antiretroviral therapy has stimulated a CD4⁺ T lymphocyte count rise. Specifically, if a patient has a CD4⁺ T lymphocyte count of 400 cells/µL while receiving antiretroviral therapy and that patient’s CD4⁺ T lymphocyte count was 50 cells/µL before antiretroviral therapy, is that patient at greater risk for developing an opportunistic infection than another patient whose current CD4⁺ T lymphocyte count is 400 cells/µL but whose nadir before antiretroviral therapy was 250 cells/µL? The data suggest that these two patients have comparable risk (i.e., the current CD4⁺ T lymphocyte count is the most important predictor of risk and the earlier nadir has only minor influence on opportunistic infection susceptibility).

Like other immunosuppressed individuals, patients with HIV infection and low CD4⁺ T lymphocyte counts require a prompt attempt to define the specific cause of their clinical syndrome. Like patients with neutropenia, fevers of unknown origin are not common. However, patients often present with specific syndromes such as pneumonia, meningitis, focal neurologic abnormalities, chorioretinitis, or diarrhea. Patients can deteriorate quickly, and the range of causative organisms is broad. Thus, as with patients with neutropenia, HIV-infected individuals need specific microbiologic and pathologic tests to determine the specific cause of their syndrome so that the appropriate therapy can be initiated, and so that unnecessary drugs can be eliminated.

In evaluating the differential diagnosis of infectious syndromes in patients with HIV (and in every other patient population as well), geography is an important part of the history. Tuberculosis is always a concern because of the extraordinary susceptibility of HIV-infected patients for developing active disease once they have been exposed.77–79 It is notable, however, that although HIV and tuberculosis overlap in many patients in much of the developing world, in the United States only 10% of cases of tuberculosis occur in HIV-infected patients.⁸⁰ In many urban settings in the United States, each pulmonary evaluation should include smears and cultures for M. tuberculosis, both to diagnose the appropriate cause of the pulmonary dysfunction and to assist in determining which respiratory precautions are appropriate. However, in the United States where only about 11,000 cases of new tuberculosis occur per year, and where 50% of cases occur in immigrants, the likelihood of tuberculosis in a U.S. native with no known exposure is quite low, in contrast to a recent immigrant from a highly endemic area.⁸⁰

In some areas of the country, such as the Ohio River Valley including Indianapolis, histoplasmosis is as common as pneumocystosis in causing diffuse pulmonary infiltrates. In the southwestern United States, coccidioidomycosis must be recognized as a cause of pulmonary infiltrates. The clinical presentations of tuberculosis, histoplasmosis, coccidioidomycosis, cryptococcosis, and toxoplasmosis can be clinically indistinguishable from PCP. Thus for HIV-infected patients with pulmonary infiltrates in an ICU, prolonged empiric therapy is discouraged in favor of vigorous efforts to establish a specific diagnosis.

Clinical Syndromes

HIV-infected patients are admitted to ICUs for major syndromes such as respiratory insufficiency, cerebral dysfunction, septic shock, hepatic or renal failure, and drug toxicities. However, patients with HIV infection also come to ICUs for routine procedures and routine postoperative care. In those situations their management ordinarily requires no extraordinary measures, with two exceptions, in addition to careful consideration of how to manage antiretroviral drugs.

First, as noted earlier, intensivists must consult with HIV specialists about management of antiretroviral drugs. The imprudent continuation of these drugs or the imprudent discontinuation of these drugs can have lifelong consequences for the patient that can be substantially avoided with proper consultation.

Second, drug interactions involving drugs used during procedures and certain antiretroviral drugs can have important clinical consequences. Many of the protease inhibitors and the non-nucleoside reverse transcriptase inhibitors that are now the backbone of antiretroviral therapy can inhibit or enhance the metabolism of drugs that depend on the cytochrome P-450 system. Thus, the half-lives of certain analgesics, sedatives, and hypnotics can be prolonged in HIV-infected patients who are taking ritonavir, for example. This pharmacokinetic effect is also relevant for a host of other therapeutic agents used in the ICU and may affect their efficacy or safety. Clinicians need to be familiar with these interactions when selecting new therapies for procedures or for clinical entities. Given how complicated these interactions are, consultation with a specialist, e.g., a pharmacologist or infectious disease specialist, is appropriate for any HIV patient admitted to the ICU when antiretroviral drugs are involved.

The therapies for specific opportunistic pathogens are summarized in Table 53.5.

Respiratory Insufficiency

Patients with HIV infection can develop severe pulmonary dysfunction because of common community-acquired pathogens such as S. pneumonia, Legionella, Mycoplasma, and Chlamydia; adenovirus; influenza; or respiratory syncytial virus, as well as other opportunistic viruses and fungi. Thus the diagnostic evaluation needs to be comprehensive, emphasizing direct smears of sputum or bronchoalveolar lavage. It is important to recognize that the clinical presentations produced by many causative agents can be similar. For instance, histoplasmosis, tuberculosis, and nonspecific interstitial pneumonitis can present identically to PCP.76,77,81,82 Thus although empiric diagnosis and empiric therapy may be reasonable as initial approaches to some patients with HIV infection and mild pneumonitis, such an approach is usually not appropriate for patients in an ICU.

Evaluation of induced sputum is the first step in the diagnostic approach to PCP. Sensitivity can be as high as 80% to 95% at many hospitals (at some institutions the yield is considerably lower).⁸³ Specificity should be 100% in an experienced laboratory. Other pathogens, including mycobacteria, fungi, and routine bacteria, can be identified in sputum as well. For intubated patients, respiratory secretions obtained by deep intratracheal suctioning are also likely to be useful, although they have not been as carefully studied as induced sputum. Should the diagnosis not be established by evaluation of sputum or intratracheal secretions, bronchoscopy should be performed. Bronchoalveolar lavage should diagnose almost 100% of cases of PCP, even if patients have already received 7 to 10 days of empiric therapy at the time of the diagnostic procedure.⁷⁶ A diagnosis of PCP is established by visualizing one or more clusters of organisms. Some laboratories are now using PCR to diagnose PCP, but this test is not standardized and is likely to be highly sensitive but not highly specific for identifying Pneumocystis as the cause of the pulmonary dysfunction.^44,84

Diagnostic criteria for other opportunistic infections are reviewed in Chapters 12 and 42. CMV merits special mention. CMV pneumonia almost never occurs in patients with HIV infection, as opposed to patients with solid organ or stem cell transplants. CMV should be considered the cause only if other causative processes have been ruled out, and there is convincing histologic or cytologic evidence. Culture of sputum or bronchoalveolar lavage for CMV does not provide useful information: in particular, patients with CD4⁺ T lymphocyte counts below 100 cells/µL will predictably have CMV present in their secretion independent of whether or not pulmonary disease is present.⁸⁵ A diagnosis of CMV pneumonia in this patient population is suggested by cytologic test and confirmed by the presence of multiple inclusion bodies in lung tissue obtained by transbronchial or open lung biopsy.

Similarly, Mycobacterium avium complex (MAC) and herpes simplex virus (HSV) can often be found in respiratory secretions of patients with HIV/AIDS by culture or by nucleic acid amplification tests, but these organisms almost never cause pneumonia in patients with HIV infection. In other patient populations they can clearly cause pneumonia, but the dearth of CMV, MAC, and HSV pneumonia in this patient population emphasizes the point that it is important to know from published literature what the clinical likelihood is for different microbial processes.

Fungal pneumonias other than PCP are generally diagnosed by direct microscopy or culture of respiratory secretions (sputum or bronchoalveolar lavage). Candida organisms almost never cause pneumonia in patients with HIV infection. The frequency of Cryptococcus, Histoplasma, Blastomyces, and Coccidioides as causes of pneumonia depends on the geographic exposure of the patient. Among these mycoses, antigen detection techniques can be useful for finding Cryptococcus and Histoplasma organisms.

Therapy of opportunistic infections is summarized in Table 53.5.⁷⁴ While awaiting a specific diagnosis, it is reasonable to initiate empiric therapy in patients ill enough to merit admission to an ICU. For patients with a CD4⁺ T lymphocyte count greater than 250 to 300 cells/µL, levofloxacin or moxifloxacin and ceftriaxone or azithromycin and ampicillin-sulbactam would be reasonable choices. For patients with CD4⁺ T lymphocyte counts below 200 to 250 cells/µL, levofloxacin or moxifloxacin plus trimethoprim-sulfamethoxazole or pentamidine plus levofloxacin or moxifloxacin would be potential regimens.

If PCP is documented, trimethoprim-sulfamethoxazole is always the drug of choice in patients who can tolerate it. Table 53.5 lists alternatives for sulfa-intolerant individuals. Regardless of which specific anti-Pneumocystis regimen is used, corticosteroid therapy is indicated for any patient who presents with an oxygen pressure (PO₂) below 70 mm Hg or an alveolar-arterial gradient higher than 30 mm Hg.^86–88 Patients with an initial PO₂ lower than 70 mm Hg are the subgroup with substantial mortality risk for whom corticosteroids have been shown to provide a survival benefit. Corticosteroids also provide more rapid resolution of pulmonary manifestations in patients who present with better pulmonary function, but survival in this population is so high that clinical trials have not been able to show survival benefit and thus corticosteroids are not conventionally recommended for patients who present with a room air PO₂ greater than 70 mm Hg. Some experts are concerned that corticosteroid use will be associated with reactivation of latent infections such as CMV or tuberculosis. However, reactivation of life-threatening infections has not been associated with this corticosteroid regimen.

How should a patient with AIDS-associated PCP be managed if there is no improvement, or if there is deterioration, after 5 to 10 days of therapy? The median time to improvement in clinical variables is 4 to 8 days; therefore, changes in therapy are probably not warranted before 5 to 10 days. At that point the accuracy of the diagnosis should be reassessed: Consideration should be given to repeat bronchoscopy with bronchoalveolar lavage or, perhaps, transbronchial biopsy to determine if CMV, fungi, mycobacteria, or a nosocomial bacterial process is present. Noninfectious processes such as congestive heart failure or tumor (e.g., Kaposi sarcoma) must also be considered. If pneumocystosis is the only causative process that can be identified, corticosteroids should be added to the regimen if they have not been already. Whether switching from one anti-Pneumocystis agent to another or whether adding a second agent is helpful has not been determined by clinical trials. Some human pneumocystosis isolates carry resistance mutations to sulfonamides, but such testing is available only in a few research centers, and the clinical significance of these mutations is unknown. Most experts add parenteral pentamidine to trimethoprim-sulfamethoxazole. Parenteral clindamycin-primaquine could be used as salvage regimens as well. Patients who have not improved after 14 to 21 days of therapy with specific chemotherapy plus corticosteroids have an exceedingly poor prognosis.

Should patients with AIDS-related PCP be intubated and provided with mechanical ventilation? The mortality rate for such patient populations was 70% to 80% in several series in the early 1980s.89,90 Since that era, supportive care has improved, and treatment modalities for concurrent infectious and noninfectious processes have become more effective. Patient selection for ventilatory support is probably also improving. Patients who have multiple active opportunistic infections, substantial weight loss, and no response to 14 days of therapy have a worse prognosis than previously ambulatory patients who develop respiratory failure during the first few days of therapy. Thus decisions about ICU support for patients with HIV infection and respiratory failure need to be individualized on the basis of a realistic assessment of prognosis, the availability of resources, and the preference of the individual patient.

As indicated earlier, the ICU is not an ideal setting for initiating antiretroviral therapy. It should be noted, however, that for PCP, early initiation of antiretroviral therapy is generally associated with increased survival and thus therapy generally should not be delayed.⁷³ However, the studies that have been done had very few patients with life-threatening manifestations of their acute opportunistic infection. Thus, for patients ill enough to be in the ICU, the initiation of antiretroviral therapy might be associated with an immune reconstitution syndrome that could make respiratory support difficult or impossible. Moreover, patients in the ICU may not be able to absorb oral antiretroviral therapy, or might have complicated drug interactions with other necessary drugs. Thus, decisions to start antiretroviral therapy in the ICU require considerable thought and do not lend themselves to simple algorithms.

Central Nervous System Dysfunction

Focal Central Nervous System Lesions

Patients with HIV infection and CD4 counts less than 100 cells/µL may present with focal motor lesions, altered mental status, or seizures. Although the differential diagnosis is extensive, the major considerations are toxoplasmosis and lymphoma.96,97

Central nervous system (CNS) toxoplasmosis may present as one or multiple lesions that represent reactivation of latent disease. Lesions are characteristically enhancing in a ringlike pattern and typically occur in the basal ganglia, but many different radiologic presentations have been documented.

All patients with CNS toxoplasmosis are IgG seropositive for toxoplasmosis using sensitive assays. However, some laboratories use less sensitive assays and thus some patients may appear to be seronegative. If lumbar puncture can be performed, a PCR for toxoplasmosis is positive in about 50% of cases, although there is no standardization of these Toxoplasma PCRs, making results variable from laboratory to laboratory.⁹⁸ The standard practice for HIV-infected patients with CD4 counts less than 200 cells/µL is to undergo an empiric trial of oral sulfadiazine plus pyrimethamine plus leucovorin. Most patients who have toxoplasmosis will show clinical and radiologic improvement within 2 weeks. If such improvement has not been documented, and the diagnosis is in doubt, a needle biopsy of the intracranial lesion should be considered.^99–101

CNS lymphoma can present identically to CNS lymphoma. A negative serum IgG for toxoplasmosis, and a negative CSF toxoplasma PCR should suggest lymphoma.¹⁰² Most patients with CNS lymphoma will have a positive CSF PCR for EBV (Epstein-Barr virus), although this test is neither 100% sensitive nor 100% specific for lymphoma. Patients who are CSF EBV PCR negative or in whom the diagnosis is uncertain may require a brain biopsy to document the lymphoma.

Therapy for lymphoma is one of several chemotherapeutic regimens. Patients continue to have a poor prognosis, but benefit from immune reconstitution with antiretroviral therapy.

Focal White Matter Lesions

In patients with HIV, JC virus encephalitis (also known as progressive multifocal encephalopathy) causes focal white matter lesions that are associated with progressive motor, sensory, and neurocognitive dysfunction.¹⁰³ A small fraction of patients can present with rapidly progressive cognitive decline.

The CT scan or MRI images of JC virus encephalitis are characteristic of this disease. Diagnosis is usually confirmed by CSF PCR for JC virus. There is no specific therapy. Patients may stabilize or improve if antiretroviral therapy results in immune reconstitution.¹⁰⁴

Diffuse Encephalopathy

Patients with HIV infection can develop significant cognitive dysfunction due to nonfocal entities including HIV encephalopathy and CMV encephalitis. HIV encephalitis characteristically progresses slowly and has ventricular enlargement and cerebral atrophy on brain imaging. The CSF is unremarkable, and there is no diagnostic CSF test.

CMV encephalitis, in contrast, is rapid in progression and has a neutrophilic pleocytosis in the CSF and a positive CSF PCR for CMV.105,106 Imaging shows either periventricular enhancement or focal nodules. Therapy is typically not successful in reversing the cerebral dysfunction. Either ganciclovir, foscarnet, or a combination of both drugs can be used in addition to the institution of antiretroviral therapy.

Diarrhea

Patients with HIV infection and CD4 counts below 100 cells/µL can develop severe diarrhea due to a wide range of opportunistic pathogens including CMV, cryptosporidiosis, or enteric bacteria such as Salmonella, Shigella, or Campylobacter. Diarrhea in HIV-infected patients can be so severe that patients can have life-threatening malabsorption, electrolyte abnormalities, and bowel perforations. Patients need a thorough stool evaluation for common pathogens such as Clostridium difficile, which is the most common cause of diarrhea in this patient population, even though it is not an opportunistic pathogen. If stool cultures are negative, testing samples for ova and parasites might be indicated if there is a history compatible with exposure.

For patients with negative evaluation of multiple stool samples, a sigmoidoscopy with biopsy is indicated to assess the presence of CMV. CMV colitis occurs exclusively in patients with CD4⁺ T lymphocyte counts lower than 100 cells/µL. CMV colitis is best diagnosed by biopsy: there is no reliable correlation with serum CMV PCR or with stool culture or PCR for CMV. The therapy of choice for CMV colitis is intravenous ganciclovir. Intravenous foscarnet is also effective but is more toxic. Oral valganciclovir is not indicated for patients with severe diarrhea but can be used as maintenance therapy once the CMV colitis is controlled. The most effective remedy for CMV colitis is immune reconstitution with antiretroviral therapy. As soon as absorption seems likely, antiretroviral therapy should be initiated.

Hypotension

Patients with HIV infection develop hypotension resulting from the same types of disorders as with non-HIV-infected individuals—sepsis from a primary infection or a wound or device (especially an intravascular access device), fluid depletion from vomiting or diarrhea, and hemorrhage from a gastrointestinal lesion are examples of common causes. The evaluation of hypotension in a patient with HIV infection must take into account factors particular to this patient population: HIV-infected patients are susceptible to opportunistic infections; they undergo many procedures that can be associated with infectious complications; and they receive an array of drugs, some of which have cardiovascular effects. Thus, evaluating hypotension in this patient population requires a comprehensive and thorough approach. A differential diagnosis of the major causes is shown in Table 53.6. Adrenal function always deserves special attention because several viral processes, fungal and mycobacterial diseases, HIV, and drugs can suppress the adrenal axis and either cause hypotension or exacerbate it.

Prevention of Opportunistic Infection

Intensivists usually focus on the management of acute processes, but if attention is not provided to opportunistic infection prevention, patients hospitalized for one process may develop a second life-threatening process given the high incidence of opportunistic infections.

Patients with HIV infection typically receive several antimicrobial agents to reduce the likelihood they will acquire opportunistic infections. Primary prophylaxis is the term used to indicate strategies that reduce the likelihood of an initial episode of a disease process. Secondary prophylaxis is the term used to indicate strategies that prevent recurrences or relapses. Chronic suppressive therapy is identical to secondary prophylaxis: This refers to regimens that are continued after the initial therapeutic course to prevent relapses.

All patients with HIV infection and CD4⁺ T lymphocyte counts below 200 cells/µL typically receive anti-Pneumocystis prophylaxis. Trimethoprim-sulfamethoxazole is the regimen of choice. Patients who actually take this drug have very few breakthroughs of PCP and receive considerable protection against toxoplasmosis and certain routine bacterial infections. Alternative regimens include monthly dapsone, weekly dapsone-pyrimethamine, or daily aerosol pentamidine. Prophylaxis against M. avium complex is recommended for patients with CD4⁺ T lymphocyte counts under 100 cells/µL; clarithromycin and azithromycin are currently the drugs of choice.⁷⁴

Some clinicians also use fluconazole or acyclovir prophylaxis to reduce the frequency of fungal and viral processes, respectively, although this is not recommended because of issues of cost, pill burden, and the emergence of resistant pathogens. Isoniazid prophylaxis is important for any patient with a tuberculin skin test that shows more than 5 mm of induration or a history of substantial recent exposure.

HIV Transmission in the Intensive Care Unit

Transmission of HIV is an issue that requires attention in the ICU.107–109 No evidence exists that HIV-infected health care professionals can infect patients, regardless of what procedure they perform, with the exception of two unusual and unexplained events. Intensivists should realize that although there are no federal policies defining how HIV-infected practitioners should be credentialed and monitored, many hospitals have policies and procedures. Some are modeled after guidelines from the Society of Healthcare Epidemiology.^108,110

HIV patients pose a risk to health care professionals.107–110 This risk can be substantially reduced by staff education, by strict monitoring for compliance with universal precautions, and by having proper equipment. Almost all HIV transmission has occurred in an occupational setting as a result of injuries involving sharp instruments (e.g., needles, scalpels). The risk of such injuries is about one case of HIV transmission per 250 injuries, but the likelihood of transmission in an individual accident depends on the amount of viremia at the time of the accident (late-stage patients generally have more circulating virus than do early-stage patients) and the nature of the accident. Most authorities recommend immediate prophylaxis if a significant injury occurs involving an HIV-infected patient. Considerable debate exists over the optimal choice of drugs and the optimal duration of therapy, but it is clear that initiating therapy within a period of hours rather than days is best. Many authorities now advocate an antiretroviral regimen for any situation when the patient and health care provider determine that therapy is appropriate, and continue that for 4 to 6 weeks. If providers have questions about appropriate medical treatment for occupational exposures, 24-hour assistance is available from the Clinicians’ Post Exposure Prophylaxis Hotline (PEPline) at 1-888-448-4911 (http://www.nccc.ucsf.edu).¹¹¹

Since surveillance of injuries began, 57 documented cases of occupational transmission of HIV infection have occurred, and 138 possible cases have been documented. More cases involve nurses or phlebotomists than physicians. No transmission in the operating suite has been documented. There have been no documented cases of transmission for over a decade.

• The risk of infection is related to the “net state of immunosuppression,” which is related to the function of the patient’s pretransplant immune status, conditioning regimens, antirejection chemotherapy, as well as severity of illness and nutritional status. Net immunosuppression cannot be measured easily: it involves the number and function of neutrophils, T lymphocytes, and B lymphocytes. Although neutrophil number is a highly reliable surrogate for susceptibility to bacterial diseases, measures of cellular immune function, which are relevant to many viral and fungal pathogens, are not reliable surrogates in terms of sensitivity or specificity for predicting disease. Thus, susceptibility to viral and fungal pathogens is more difficult to predict.

• The risk of noninfectious complications is dependent on the organ transplanted and the immunosuppressive regimen used.

• Distinguishing infectious from noninfectious complications can be challenging, emphasizing the need for a broad consideration of causes of fever, hypotension, or organ-related syndromes, and the institution of prompt diagnostic tests that focus on both infectious and noninfectious causes.

• With solid organ transplants or stem cell transplants, the donor is a source of infection if the donor had an unrecognized transmissible infection at the time of organ donation, or if the donor had a latent infection that was transmitted with the donated organ.

• With solid organ transplants, complications of the surgical procedure must be considered when postoperative complications occur.

• Organ transplant recipients characteristically have extensive contact with health care environments and may have been receiving antimicrobial prophylaxis: these factors influence the spectrum of likely causative organisms.

• Diagnostic studies for infections rely more and more on molecular testing, although tissue biopsy and cultures of blood and suspicious body fluids and anatomic areas remain important. Tissue biopsies can be especially useful for differentiating infectious and noninfectious causes.

• Drug levels and microbial markers of infection must be carefully monitored to enhance the likelihood of effective therapy and minimize the likelihood of drug-related toxicities.

Diagnostic Approach

The management of organ transplant recipients focuses on algorithmic monitoring of neutrophil and lymphocyte number, organ function, drug levels, prophylactic drugs, and microbiologic parameters. Most patients will be managed by standard practice protocols that routinely order tests at intervals appropriate for the time period after transplant and for the patient’s clinical status.

Programs often have specific targets for drug levels for immunosuppressive drugs. Given the drug interactions that occur in the ICU due to polypharmacy, and other factors altering pharmacokinetics, drugs levels must be monitored closely.

Infectious Complications of Transplantation

Noninfectious Complications in Human Stem Cell Transplant Recipients