Infection and Diabetes

Deborah E. Sentochnik

George Eliopoulos

It is commonly believed that the incidence of infection is higher in persons with diabetes mellitus and that such infections result in complications and death more frequently than would be anticipated in otherwise healthy individuals (1,2). Older studies, upon which much of this information is based, focus particularly on infections of the urinary tract, the respiratory tree, and the extremities and derive their data from autopsy cases. However, in these studies the degree to which infection at these sites actually contributed to the cause of death is frequently not clear, and control groups are typically lacking. More recent studies, while documenting excess mortality among patients with diabetes, have ascribed this largely to cardiovascular disease rather than to uncontrolled infection (3,4).

In diabetes mellitus, a number of factors greatly complicate efforts to assess risk of infection and resulting complications. The most basic is the problem of determining an appropriate estimate of the population at risk, which may be difficult to obtain for diabetes and is rarely if ever presented. Furthermore, in diabetes, as in other chronic diseases in which the natural history for any individual may span decades, historical controls are of limited utility, given the expected improvements in the general health of a population, the development of more effective diagnostic techniques, earlier medical intervention, and the availability of expanded therapeutic options, including more active and better tolerated antimicrobial agents. A number of variables, including duration of illness, severity of noninfectious complications, concurrent illnesses, level of glucose control, and even degree of medical supervision, result in a very heterogeneous group of individuals at risk even within a more narrowly defined time frame. Finally, some infections that may be particular to diabetics, such as emphysematous cholecystitis, are so uncommon that information regarding risk factors and management options is limited.

Despite these limitations, much is known both about those uncommon infections that occur predominantly in patients with diabetes mellitus and about the more common infections that, while not restricted to those with diabetes, will often complicate the general management of this group of patients. To acknowledge such limitations in advance is to underscore the need for careful individualization in the approach to diagnosis and therapy for any diabetic patient with suspected or proven infection.

DIABETES AND THE IMMUNE SYSTEM

Defining altered host responses in diabetes has long been hampered both by the complexity of the systems in question and by
the rather limited availability of techniques to study these responses. in vivo the various arms of the immune system are highly dynamic and interdependent. It is thus simplistic to study any single component of it in vacuo. Historically, however, methodologic constraints did in fact limit such studies to individual aspects of host defenses, such as leukocyte adherence or phagocytosis, exclusive of other components of the system. Even more recent has been the increase in the appreciation of complex interactions, not only among the various cellular elements of the immune system itself but also among the elements of the immune system and other body components such as the vascular endothelium. As the approaches to the investigation of these issues evolve, it becomes increasingly more difficult to compare studies. It is for these reasons that, even after decades of investigation, questions remain about whether diabetes itself results in specific immunologic defects and how such defects might predispose to infection (5).

Function of Polymorphonuclear Leukocytes

MOBILIZATION AND CHEMOTAXIS

Using the Rebuck skin window technique, which was to remain the standard for many years, Perillie et al. (6) studied polymorphonuclear leukocyte (PMN) chemotaxis in ten patients with well-controlled diabetes, six patients with diabetic ketoacidosis, four patients with nondiabetic uremic acidosis, and ten healthy controls. An abrasion was created on the volar forearm, and sterile coverslips were serially applied over the next several hours. Mobilization of PMNs to the area of inflammation was graded by microscopic examination of the coverslip after staining. The PMNs of all acidotic patients had a diminished early (24 hour) response. This response time became normal in the four patients with diabetes whose acidosis was corrected. In an ambitious analysis of leukocyte function, Brayton et al. (7) examined chemotaxis with use of a modified Rebuck technique in 18 patients with fairly well-controlled diabetes, five of whom were acidotic at the time of study. At 2 and 4 hours, mobilization of the PMNs of all diabetic subjects was diminished. However, at later time points, no differences were seen between the mobilization responses of the PMNs of diabetic subjects and those of healthy controls. The PMNs of acidotic uremic patients had a chemotactic response similar to that of controls at all time points.

Mowat and Baum (8), in an in vitro study using a modified tissue-culture chamber, studied chemotaxis of PMNs from 31 diabetic patients with various degrees of glycemic control. A chemotactic index was derived by comparing the original number of PMNs with the number that had completely crossed a filter barrier in response to chemoattractants. The PMNs of all patients with diabetes had a lower chemotactic index (i.e., diminished response), without any correlation between chemotactic index and type of therapy or fasting blood glucose levels. Incubation of the PMNs from 11 controls with glucose at concentrations of 100 to 900 mg/dL did not change the chemotactic index of the PMNs. Incubation of the PMNs of diabetic patients with insulin at concentrations of 10 to 100 μU/mL improved the chemotactic index of PMNs of the diabetic patients if glucose was also present. Molenaar et al. (9) used a bacterial factor from Escherichia coli as a chemoattractant and found a lower chemotactic index for the PMNs of 52 first-degree relatives of 15 patients with diabetes as compared with the chemotactic index for the PMNs of controls. Not all the subjects with diabetes had PMNs with a depressed chemotactic index, but the average chemotactic index was lower than that for the PMNs of their relatives. A later study (10) that used a similar technique (Boyden modified chamber) with cells washed free of plasma found no difference in the chemotactic index of the PMNs of control subjects and those of patients with insulin-dependent diabetes mellitus (IDDM) with various degrees of glucose control or duration of disease.

Shortly after the above studies were conducted, there was a shift to the subagarose technique for assaying chemotaxis. This technique yielded more reproducible results and corrected for chemokinetic movement. Measurements are made of the migration from a center well in an agarose plate toward a chemoattractant (zymogen-activated plasma) and of random migration toward a control well. With this technique, the average chemotactic index of the PMNs of 58 patients with diabetes was found to be depressed (11). Chemically defined chemotactic activity of PMNs under agarose was investigated by Naghibi et al. (12) for 26 patients receiving oral hyperglycemic agents, daily insulin injections, or continuous insulin infusion before an intensive control regimen and in 11 of these patients after institution of the intensive regimen. Chemotaxis of the PMNs of all groups was comparable to that of control PMNs.

PHAGOCYTOSIS

Bybee and Rogers (13) studied leukocyte phagocytosis in 31 patients with well-controlled diabetes, seven patients with diabetic acidosis, and a control group. Washed PMNs and an equal number of Staphylococcus aureus or Staphylococcus epidermidis were incubated together for 60 minutes in 10% human serum. Phagocytosis was considered to be present if at least one bacterium was ingested. There was no quantitation of the number of organisms engulfed per cell. Only the PMNs of ketotic diabetic patients were found to exhibit diminished phagocytosis. This defect was corrected if acidosis was reversed but not if the cells were incubated with normal serum. Control PMNs functioned normally if they were incubated with serum from acidotic diabetic patients. However, serum factors may have been diluted out, given that a 10% concentration of serum was used.

Bagdade et al. (14) used a similar system, but with 90% serum, to examine phagocytosis of Streptococcus pneumoniae type 25 by leukocytes from eight patients who were not acidotic but had poorly controlled diabetes. Decreased phagocytosis was especially notable when fasting blood glucose levels were higher than 250 mg/dL. After the patients’ glucose levels were controlled, phagocytic activity improved but did not attain control values. In contrast to the findings of Bybee and Rogers (13), the activity of control PMNs was diminished when the cells were incubated with serum from patients with diabetes, whereas the activity of PMNs from patients with diabetes was increased when the cells were incubated with normal serum. The work of Rayfield et al. (15) supported the possibility of an opsonization defect affecting PMNs of individuals with diabetes. Normal PMNs had decreased uptake of radiolabeled E. coli or S. aureus in the presence of serum from patients with diabetes.

Using the lysostaphin assay technique, which allows differentiation between phagocytosis and intracellular killing, Tan et al. (16) demonstrated a defect in phagocytosis of S. aureus by the PMNs of 31 patients with adult-onset diabetes. The presence of the defect showed no correlation with level of glycemic control or history of recurrent infections. The addition of normal serum had no effect. Using shorter observation periods, Nolan et al. (17) found that PMNs from 17 patients with poorly controlled diabetes ingested a smaller proportion of an inoculum of 10⁶ S. aureus after 20 minutes (the interval during which most engulfment takes place under normal physiologic conditions) than did control PMNs. This difference vanished at 60 minutes. Davidson et al. (18) measured engulfment of Candida guilliermondii over a 45-minute period by PMNs from 11 patients with moderately well-controlled diabetes. The ratio of white cells to organisms was such that 90% of the control cells
would have ingested at least one yeast cell in 30 minutes. Phagocytosis was diminished in the PMNs of the patients with diabetes, regardless of levels of glycosylated hemoglobin 1C (HbA_1c). A defect in opsonization was suggested, since preopsonized yeast particles added to serum from diabetic patients with PMNs from control subjects were engulfed at normal levels. Alexiewicz et al. (19) demonstrated impaired phagocytic ability of PMNs from patients with newly diagnosed, non-IDDM, which correlated inversely with fasting serum glucose level. Both phagocytosis and glucose control improved significantly after 3 months of therapy with glyburide. The authors also demonstrated an inverse relationship between cytosolic calcium concentrations in PMNs and phagocytic function and postulated that the functional defect may relate to the high calcium levels.

ADHERENCE

Comparatively few papers have specifically addressed the question of adherence by the leukocytes of patients with diabetes. Peterson et al. (20) found that the PMNs of six of seven patients with poorly controlled diabetes exhibited impaired adherence to a glass-wool column. Adherence improved 1 to 2 months later when glycemic control had improved. However, no control patients were examined. Bagdade et al. (21) showed an enhancement of adherence of PMNs to a nylon-fiber column following an improvement in the control of blood glucose levels. Adherence increased from 53% to 74% of control values. In another study, Bagdade and Walters (22) demonstrated a direct relationship between degree of glucose control and PMN adherence.

Andersen et al. (23) pointed the study of adhesion in a dramatic new direction by devising a more physiologic system. Noting that vascular endothelium is not a passive participant in the inflammation cascade, they examined the ability of PMNs from 26 patients with diabetes and from age-matched controls to bind to bovine aortic endothelium. The PMNs of 60% of the patients with diabetes had severely depressed function that did not correlate with HbA_1c levels. PMN-PMN aggregation was not defective. No quantitative defect in fibronectin was seen, but a qualitative defect could not be excluded.

BACTERICIDAL ACTIVITY

Early studies, such as that of Dziatkowiak et al. (24), compared the number of live S. aureus bacteria in a granulocyte with the total number engulfed to calculate the proportion of organisms killed. Several studies (16) demonstrated diminished killing by the PMNs of patients with diabetes while others did not. Repine et al. (25) took a more quantitative and functional approach. Instead of using a single low ratio of bacteria to PMNs, they used five different ratios (1:1 to 100:1). Study patients included infected and noninfected individuals with and without diabetes. Cells were incubated with S. aureus for 1 hour, after which colonies were counted. The rates of intracellular killing of bacteria by PMNs from uninfected controls and by PMNs of persons with well-controlled diabetes were comparable. PMNs from uninfected patients with poorly controlled diabetes functioned less well, especially when the higher ratios of bacteria to white cells were used. Although the functioning of PMNs from infected patients with well-controlled diabetes was on a par with the functioning of those from uninfected controls, the PMNs from the infected patients with diabetes did not display the increase in killing activity seen in the PMNs of infected patients without diabetes. The bactericidal function of PMNs from infected patients with poorly controlled diabetes was the lowest of all the groups. Naghibi et al. (12), using a single low ratio of bacteria to PMNs, found depressed bactericidal function of the PMNs of patients against Pseudomonas aeruginosa, but they did not study any infected patients. Serum from patients with diabetes had an inhibitory effect on PMNs from both normal controls and diabetic subjects before and after intensive glucose management. Bactericidal activity of PMNs from the patients with diabetes remained depressed even after intensive management.

Stimulated PMNs display a burst of oxidative metabolism that produces superoxide anions and other oxygen-derived species implicated in bacterial killing. These reactions produce chemiluminescence, a sensitive indicator of oxidative metabolism that correlates with antimicrobial activity. Shah et al. (26) looked at the production of superoxide anion and chemiluminescence of PMNs from patients with diabetes, examining the cells in both the resting state and in response to soluble and particulate stimuli. In resting PMNs from patients with diabetes, the chemiluminescence of cells placed in serum from patients with diabetes was comparable to that of cells placed in control serum. Superoxide production was higher in autologous serum from patients with diabetes than in normal serum. The significance of these findings taken together was unclear. When stimulated, the PMNs from patients with diabetes showed a blunted response with regard to both superoxide production and chemiluminescence. Cross-incubation serum studies effected no change, suggesting that an intracellular defect rather than an inhibitory serum factor might have been present. The precise role of defective in vitro bactericidal activity of PMNs, associated with impaired superoxide production, as a predisposing factor for infection is uncertain, however. Such defects can be demonstrated in patients with diabetes who have not been subject to recurrent or particularly serious infections (27).

Sato et al. (28) reported improvement in chemoluminescence measurements related to O₂^– and OCl^– production by PMNs of patients with diabetes after 4 weeks of therapy with an aldose reductase inhibitor, although levels did not reach those of PMNs from healthy controls. This observation could not be attributed to improvement in glucose control, as both the postprandial glucose and HbA_1c levels were unchanged. It has been suggested that advanced glycation products associated with diabetes may bind to specific motifs common to lactoferrin, lysozyme, and other antimicrobial proteins found in PMNs and interfere with the antimicrobial function of these host defense molecules (29).

The PMNs of patients with diabetes have shown a decreased prostaglandin E and thromboxane B₂ response to stimulation by zymosan or killed S. aureus (30). The synthesis and release of leukotriene B₄ by these cells also were diminished compared with those of the PMNs of sex- and age-matched controls (31). The significance of these findings is not known.

Monocyte Function in Diabetes

Geisler et al. (32) found a decrease in the total number of circulating monocytes in 14 patients with diabetes. These cells displayed diminished phagocytosis of Candida albicans but not of latex particles or sheep red blood cells. Glass et al. (33) proposed that monocytes from patients with diabetes have a diminished activity of “lectinlike” receptors necessary for the recognition of cell-wall components of microorganisms. Attachment of S. epidermidis to these monocytes was impaired, but attachment to coated sheep red blood cells, which are recognized by the Fc receptor, was normal. It could not be assessed whether the proportion of monocytes with the lectinlike receptors was reduced, each receptor had a lower affinity, or each monocyte had fewer receptors. Katz et al. (34) described subpopulations of monocytes with a reduced ability to phagocytose Listeria monocytogenes.
Impaired monocyte chemotaxis has also been reported (35). Monocytes from patients with diabetes have been found to exhibit increased adhesion to fibronectin. Although this property may play a role in the genesis of atherosclerosis (36), its relationship to antimicrobial function is not known. The metabolic activity in response to the ingestion of zymosan particles by the monocytes from patients with diabetes is increased, as reflected by higher levels of chemiluminescence, superoxide production, and hexose monophosphate shunt activity than those exhibited by control monocytes (37). The consequences of this increased activity have not been evaluated. Studies of monocytes from patients with diabetes report upregulated secretion of inflammatory mediators such as tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), and prostaglandin E₂ (38).

Cell-Mediated Immunity

Reports of defects of cell-mediated immunity in vitro in patients with diabetes abound. Unfortunately, results are piecemeal both because of the complex interrelationships involved in the cell-mediated immune system and because of the evolution of study techniques over time.

MacCuish et al. (39) found that lymphocyte transformation in response to the mitogen phytohemagglutinin (PHA) was diminished in patients with poorly controlled diabetes. Meanwhile, Casey et al. (40,41) determined that the transformation of lymphocytes of patients with diabetes in response to PHA was normal regardless of the patient’s glycemic control but that the response to a staphylococcal antigen was decreased. These authors did not mention any ketotic patients. In a study by Speert and Silva (42), the lymphocytes of children with diabetic ketoacidosis had a decreased mitogenic response that reverted to normal when metabolic derangements were corrected. There may be a diminished release of migration-inhibition factor by T lymphocytes from patients with diabetes (43,44). T-lymphocyte subsets have been studied with regard to the possibility of an autoimmune basis of diabetes. While alterations of CD4-to-CD8 lymphocyte ratios during the evolution and progression of diabetes have been noted (45,46,47), no relationship of these changes to infection has been detected. No agreement has been reached as to whether the number and function of T and B cells in patients with diabetes is increased, decreased, or normal as compared with those in controls (10).

An acquired defect in the production of IL-2 by T cells has been demonstrated in patients with IDDM (48), as have increased levels of receptors for this cytokine (49). Decreased responsiveness to interferon of natural killer cells from patients with IDDM has been observed (50). The implications of these abnormalities with regard to defense against infection remain speculative at the present time.

Miscellaneous Factors

Abnormalities in the microvascular circulation of individuals with diabetes may result in decreased tissue perfusion (51). While it is intuitively understandable that such abnormalities might facilitate the acquisition of infection and impair response to therapy, it is unclear what role microvascular defects actually play in the pathogenesis of infections relatively specific for patients with diabetes, such as mucormycosis, malignant external otitis, and emphysematous cholecystitis. Reviews of these topics often mention the arteriolar narrowing seen on pathologic examination, but comparisons with control specimens have not been reported. It appears that the white blood cells of patients with diabetes may play a role in producing damage to the capillary and venular endothelium (52,53).

INFECTIONS STRONGLY ASSOCIATED WITH DIABETES

Mucormycosis

THE ORGANISM AND HOST RESPONSE

The term mucormycosis connotes a variety of infections caused by fungi belonging to the order Mucorales, members of the class Zygomycetes (54,55). Zygomycosis and phycomycosis are synonyms that have been rendered obsolete by ongoing reclassification. Rhizopus species (especially Rhizopus oryzae and Rhizopus arrhizus) are the most commonly isolated pathogens, followed by Mucor organisms. Cunninghamella (56) and other species have also been found to cause disease. These molds produce large, thick-walled, nonseptate hyphae that branch at more-or-less right angles. Ubiquitous in the environment, these organisms are most often found in decaying matter. Humans commonly inhale the spores, which have a low virulence potential. The ability of the spores to germinate successfully is dependent on specific host factors. Most information regarding pathogenesis comes from studies in rabbit and mouse models. After inhalation of spores, normal animals do not become ill, whereas diabetic animals develop a rapidly progressive pulmonary disease (57). Alveolar macrophages from normal mice, but not those from diabetic mice, will ingest the spores and inhibit their germination into the invasive hyphal forms. Experimentally, neither hyperglycemia nor metabolic acidosis alone is sufficient to permit infection (58), despite the propensity for the development of certain mucormycosis syndromes among acidotic patients with diabetes. In contrast to normal human serum, the serum of patients with diabetes with ketoacidosis does not inhibit the growth of R. oryzae (59). It has been proposed that acidosis disrupts the ability of transferrin to bind iron, a deficiency that results in the release of free iron into the serum and perhaps to an interference in the host defenses against Rhizopus (60), an iron-requiring organism (61). Credence has been lent to the proposed role of iron regulation in host defense by the increasing recognition of the occurrence of deferoxamine-associated mucormycosis in patients undergoing dialysis. In this situation, it appears that Rhizopus organisms may be able to use the iron mobilized by the chelating agent, a capability that has been demonstrated in other organisms (61).

CLINICAL SYNDROMES

Mucormycosis may present as a rhinocerebral, pulmonary, cutaneous, gastrointestinal, or disseminated form of the disease. Rhinocerebral mucormycosis was first recognized more than 50 years ago (62). It occurs primarily in persons with diabetes, although other immunocompromised patients may be affected, and is one of the most fulminant forms of fungal disease affecting that population. The typical patient presents with ketoacidosis. Fungal elements gain entry through the nasopharynx, where tissue invasion may result in nasal discharge that may be tinged with blood. Close inspection of the infected region may reveal necrotic areas with black eschar involving the nasal mucosa or hard palate. The patient is likely to have a fever and to remain lethargic even after metabolic derangements have been corrected. Commonly, by the time the diagnosis is suspected, headache and/or facial pain already reflect extension of the process into the paranasal sinuses and possibly into the orbit of the eye. Occasional patients present with a dramatic, rapid onset of ocular proptosis and vision loss caused by invasion of the orbit. Rapid progression of clinical findings is the result of invasion of blood vessels, with vascular occlusion and subsequent necrosis of tissues dependent on the affected vessels. Progression of thrombosis can include the cavernous sinus (63)
and the internal carotid artery (64). Invasion of the brain results in meningoencephalitis and/or abscess formation with deterioration of neurologic function.

Progression usually occurs over a matter of hours to days and is invariably fatal if the patient is not treated early. Diagnosis must be established by demonstration of characteristic invasion of tissues by hyphal elements in biopsy specimens. A careful search for evidence of vascular invasion must be conducted. Culture results will often be negative. Spinal fluid findings are nonspecific (65). Sinus films may show mucosal thickening and clouding with some spotty destruction of the orbit (66). Computed tomography (CT) scans, with special orbital views, can help define the extent of involvement, although the extent estimated by this method may underestimate the actual extent of involvement determined at surgery (67). Angiography has been used to define the degree of large-vessel involvement. Magnetic resonance imaging offers the potential to provide useful information on the extent of infection, including assessment of soft tissue and vascular structures. Extension beyond the orbit carries a poor prognosis.

Aggressive surgical management is mandatory in all situations; radical debridement sometimes necessitates orbital exenteration. Repeated debridement may be necessary. Concomitantly, underlying metabolic disorders must be addressed. Amphotericin B remains the standard antifungal therapy and must be used in conjunction with surgery (54,68,69). Aggressive antifungal therapy should be used, and although the exact duration of treatment and the total dosage administered have not been well defined, it is reasonable to aim for a total dosage of at least 2 g. The currently approved azole antifungal agents have no defined role in therapy (70). Repeated imaging is necessary to follow the response to treatment, and repeated biopsy may be necessary (71). With optimal therapy, mortality remains at 50%, despite some reports of more encouraging results (72). The nature of the almost universal neurologic residua depends on the anatomic structures that have already been compromised by the time the infection is brought under control.

The remaining forms of mucormycosis do not have a predilection for a specific host, although the pulmonary disease may have a distinctive presentation in patients with diabetes. In contrast to the fatal pneumonia and extensive thrombosis that is typical in immunocompromised patients, diabetic patients have been noted to develop endobronchial and large-airway lesions that may follow a less fulminant course. The main complication of pulmonary mucormycosis is massive hemoptysis. Lesions may respond to aggressive local resection in combination with intravenous therapy with amphotericin B (73,74,75,76).

Malignant External Otitis

In 1968, Chandler (77) described a series of patients with a disorder termed malignant external otitis (MEO). Occasionally referred to as progressive, invasive, or necrotizing external otitis, these names attest to the destructive nature of the process. These terms connote a slowly progressive cellulitis that begins in the soft tissues of the external auditory canal. As it penetrates more deeply into the subcutaneous tissue, this process may spread via the fissures of Santorini (clefts in the cartilaginous floor of the external canal) into the mastoid air cells. Access to the temporal bone (osteomyelitis) occurs through the cartilaginous/osseous junction in the outer ear canal (78). The great majority of cases are due to P. aeruginosa even though this is not a normal colonizer of the ear in any patient group (79). Most patients are elderly, and 75% to 90% have diabetes. There does not appear to be a distinct relationship between MEO and ketoacidosis or the magnitude of hyperglycemia (80).

Typically, patients will present with a 2- to several-week history of external otitis unresponsive to local therapy. The evolution into MEO, with localized invasion of soft tissue and osseous structures, is heralded by unrelenting, severe otalgia often accompanied by purulent discharge (81). It is unusual for patients to appear systemically ill or to have a fever and an elevated white blood cell count. One of the hallmarks of MEO that distinguishes it from simple external otitis is the finding of granulation tissue, usually at the junction of the cartilaginous and osseous portion of the canal, in more than 90% of cases. A swollen, reddened, moist-appearing canal is usual. The tympanic membrane is normal in the rare cases in which it can be visualized. The extent to which the process has progressed toward the base of the skull is reflected clinically by progressive involvement of cranial nerves. The facial nerve may be impaired at presentation in up to 50% of cases (78,82), the result of swelling of the soft tissue surrounding the styloid foramen where the nerve exits the skull or of direct invasion of the bone at the foramen itself. The function of cranial nerves IX, X, and XI can be affected next as the jugular foramen becomes involved (82,83). Finally, the hypoglossal canal can be destroyed. In the most extreme cases, contralateral cranial nerves are compromised as the destructive process erodes the base of the skull. Meningitis may result by extension into the subarachnoid space.

Plain films of the ear canal are of limited sensitivity and specificity for this condition. While technetium scanning is highly sensitive, it is of low specificity, as is gallium scanning. CT scanning, with special views, is a useful imaging modality (84), although tumors cannot be reliably distinguished from MEO. Magnetic resonance imaging provides excellent detail and is, perhaps, the most widely used modality for the initial assessment of patients suspected of having this disorder.

In Chandler’s original series (77), the mortality rate from MEO was 50%. A 1981 review of this entity cited a mortality rate of 20% (80). Cure rates since 1985 may have reached 90% (85). The improvement in survival may be attributable in part to the development of more effective antipseudomonal antibiotics. As important, however, is the likelihood that increased awareness by physicians of this entity has led to earlier recognition of affected patients and thus to the possible control of infection with antibiotic therapy and local debridement and to a diminished need for the more radical surgical procedures necessitated by more advanced disease.

Standard antibiotic regimens for treatment of MEO due to P. aeruginosa previously consisted of an antipseudomonal penicillin plus an aminoglycoside for a prolonged period (4 to 6 weeks or more) (82,86). With the advent of newer, poten-tially less-toxic antipseudomonal agents, new regimens have evolved. Monotherapy with the antipseudomonal third-generation cephalosporin, ceftazidime, for 4 to 6 weeks, much of it as home therapy, realized a rate of favorable response of 92% in a series of 20 patients, 30% of whom presented with ipsilateral facial palsy (87). Follow-up at 1 year showed no recurrences. While the results are encouraging, it is of note that two of the patients in this series did not even need debridement, indicating that some presented with the extreme of mild disease. In another study, 21 of 23 patients with MEO were successfully treated with 6 weeks of oral therapy with ciprofloxacin (88). However, only 65% of the patients in this study had diabetes. At present, therapy with an antipseudomonal fluoroquinolone—which may permit oral administration—appears to be an acceptable approach in most patients with a susceptible organism. However, close medical supervision is necessary, as repeated debridement may be required (89). While emergence of resistance has not been a frequent problem in studies using
monotherapy (or combination therapy), this possibility should be considered in patients for whom therapy fails.

Emphysematous Pyelonephritis

Severe bacterial urinary tract infection in the patient with diabetes may result in emphysematous pyelonephritis. The precise definition of this rare entity varies among authors. The most rigorous definition includes a requirement for the presence of gas within the renal parenchyma, which may enter the peri-nephric space by extension. Gas can occur in the calices, collecting system, or bladder, but if it is limited to these sites, these entities are distinct from true emphysematous pyelonephritis and carry different prognoses. The differential diagnosis includes iatrogenic manipulation with introduction of air and fistula formation arising from the digestive system. The first reference to emphysematous pyelonephritis is usually cited as 1898 (90), although in retrospect, it appears that this patient probably had gas only in the collecting system. A number of cases have been reported, but the disorder is rare (91,92,93,94,95). Between 85% and 100% of patients have had diabetes mellitus. Up to 40% of patients may have had concomitant obstruction, which is present in almost all nondiabetic patients with emphysematous pyelonephritis. At presentation, it is not possible clinically to distinguish emphysematous from uncomplicated acute pyelonephritis unless gas has spread beyond the bounds of perirenal tissues to cause subcutaneous crepitation. Pneumaturia is distinctly unusual.

Plain radiographs (96) ultrasound, or abdominal CT scans will reveal gas within the renal parenchyma. A plain x-ray film may show a mottled renal parenchyma with gas bubbles, often in a radial distribution. As the infection advances, gas may be seen outside the renal cortex either outlining the kidney or forming thin crescents, although this classic finding is unusual. Abdominal CT scanning is especially useful for documenting the spread of gas beyond the Gerota fascia. About 10% of patients have bilateral involvement (93). Radiologic evaluation should be undertaken if a suspected case of acute pyelonephritis does not respond to adequate antibiotic therapy within a few days. CT scans may also uncover previously unsuspected abscesses.

The usual causative organisms are typical urinary patho-gens, including E. coli, Klebsiella pneumoniae, Proteus mirabilis, and Enterobacter aerogenes. There are case reports implicating other organisms, including Candida species (97); typically, anaerobes are not found. It is not clear why the microbes involved produce gas in a specific clinical situation. The presence of glycosuria, while providing a substrate for production of gas by fermentation, is clearly not the sole factor, given the rarity of emphysematous pyelonephritis and the frequency of glycosuria.

While it is clear that therapy with antibiotics directed at the offending organisms and management of underlying diabetes are imperative, the role of surgical intervention is much less well defined. The lack of prospective, controlled studies is understandable given the rarity of the syndrome. Obstruction should always be sought and relieved as necessary. Some authors suggest following the resolution of gas radiographically and proceeding with nephrectomy only if the patient fails to respond to appropriate antibiotics (93). Others have recommended total nephrectomy as soon as the diagnosis is made, citing the possibility of rapid clinical deterioration in some patients (92). A reasonable approach in an otherwise relatively stable patient might be to try medical management with potent antibiotics, relieve any obstruction, undertake percutaneous drainage as appropriate, and consider nephrectomy if clinical improvement does not occur. Nephrectomy may be especially appropriate if the affected kidney is shown to be nonfunctioning, as occurs in about one half of the patients (94). However, clinical decisions are often not straightforward, as in the case of bilateral emphysematous pyelonephritis or disease in a solitary functioning kidney.

The duration of antibiotic therapy is not addressed in various reviews, but a several-week course would appear prudent. Reported mortality rates range from 10% to 40% (92,93,98). At autopsy, 50% of 42 patients had severe acute and chronic necrotizing pyelonephritis with multiple cortical abscesses, 20% had papillary necrosis, and 20% had intrarenal vascular thrombi (95). In one fourth of the cases, the kidney could not even be identified. Currently, this rare entity may carry a better prognosis if a high index of clinical suspicion is maintained to ensure early imaging studies.

Emphysematous Cholecystitis

Emphysematous cholecystitis is a rare complication of acute cholecystitis in which air is found in the lumen and wall of the gallbladder, with possible extension to the pericholecystic space. It was first diagnosed at surgery in 1908, with the first preoperative diagnosis made by abdominal roentgenography in 1931 (99). In 1975, Mentzer et al. (100) published their comprehensive study, which encompassed the 161 cases reported to date in the literature and three cases of their own. More than one third of the patients in this series had diabetes. In contrast to acute cholecystitis, for which almost three fourths of the patients are female, males account for three fourths of the cases of emphysematous cholecystitis. Perforation and gangrene of the gallbladder are 30 times and three times more common, respectively, and mortality is ten times higher in patients younger than 60 years old but only twice as high for those older than 60 years than in patients with acute cholecystitis. Diabetes does not appear to be an independent factor relative to a worse prognosis. Cholelithiasis is present about one half of the time.

Patients present with pain in the right upper quadrant, nausea and vomiting, and fever. During the next 48 hours, gas develops within the gallbladder lumen and wall, with progression into the surrounding tissues in the next 48 hours (101). Even during this time, one cannot reliably distinguish acute uncomplicated cholecystitis from emphysematous cholecystitis at the bedside. Diagnosis is established by radiographic documentation of gas in the aforementioned areas. On plain films, one may initially see a globular shadow representing the air-filled gallbladder. Soon afterwards, intramural—usually submucosal—air may be visualized (99). Air may sometimes be seen only in the biliary radicles (102). Air under the diaphragm suggests perforation and is a poor prognostic sign (99). Differential diagnosis of these radiographic findings includes the presence of enterovesicular fistula. Ultrasonography has proven a useful tool, although plain films must still be obtained to rule out a porcelain gallbladder or a heavily calcified stone as the source of high-density echoes (103,104). CT is another useful modality in making the diagnosis.

Given the high complication rate of emphysematous cholecystitis, it is important to consider this diagnosis early in the patient with diabetes who is suspected of having acute cholecystitis. It has been recommended that cholecystectomy be performed as soon as the diagnosis is established (101). Typically, a crepitant gangrenous gallbladder is found at surgery. Cultures are positive 50% to 90% of the time, a frequency much higher than that for simple acute cholecystitis. Clostridium perfringens is isolated from 25% to 50% of positive cultures, along with more typical enteric organisms such as E. coli. Antibiotic coverage
should include anaerobes as well as both gram-positive and gram-negative facultative organisms.

The issue of elective cholecystectomy for asymptomatic gallstones in patients with diabetes has been a source of controversy. Because of the perceived increased risk of mortality or severe complications among diabetic patients with acute cholecystitis, for many years it was recommended that those with cholelithiasis undergo prophylactic cholecystectomy. However, complications appear to be uncommon when diabetic patients with asymptomatic stones are followed without surgery. Del Favero et al. (105) followed 47 non-insulin-requiring diabetic patients with asymptomatic gallstones over 5 years; only 15% developed any symptoms, and only two patients developed acute cholecystitis or jaundice. A retrospective controlled study suggested that diabetes alone was not correlated with a higher mortality rate during an episode of acute cholecystitis (106). In a decision-analysis model, it was found—under a broad range of assumptions—that, when managed expectantly (i.e., observed without surgery), diabetic patients with asymptomatic gallstones did no worse than nondiabetic patients (107). Other work suggests a higher rate of infectious complications following cholecystectomy for acute cholecystitis among diabetic than nondiabetic patients (108). In treatment of acute cholecystitis, the trend has been toward expeditious (within 24 hours) surgery (101). The outcome of this approach, in contrast to that of surgery following a “cooling-off” period, has not been rigorously examined in relation to patients with diabetes. Nevertheless, while severe complications are infrequent when diabetic patients with asymptomatic gallstones are observed expectantly, laparoscopic techniques have also reduced the morbidity of cholecystectomy. Therefore, decisions regarding benefits of elective cholecystectomy can be made on a case-by-case basis (109).

INFECTIONS CAUSED BY THERAPEUTIC INTERVENTIONS

Only gold members can continue reading. Log In or Register to continue