A high level of awareness is required for patients with EPN due to the life-threatening nature of the associated septic complications. Treatment of EPN initially involves fluid and electrolyte resuscitation, antibiotic therapy, glycemic control, and relief of urinary tract obstruction where appropriate. Further management consists of either conservative medical treatment with antibiotics, emergency nephrectomy with medical therapy, or PCD with medical therapy. The treatment of EPN has evolved over the years from invasive surgery to more conservative approaches.
Some studies advocate nephrectomy for patients with EPN, in combination with antibiotics, fluids, and optimization of blood glucose and electrolytes [3, 8]. However, the mortality rate in a series by Ahlering et al. advocating emergency nephrectomy was 42% . Some specialists have proposed that conservative treatment or PCD should only be used for select cases such as localized EPN, bilateral disease, EPN in a solitary kidney, or if the patient cannot tolerate general anesthesia . Medical treatment with antibiotics alone carries the highest risk of mortality . However, in the current study, 10 patients received antibiotics alone, and the mortality rate was 20% (2 patients died). Most of the cases in the current study were limited EPN (class 1 & 2) and were comparatively healthy. Six patients had class 1 EPN, 3 had class 2, and 1 had class 3A with a small air pocket. This may account for the superior outcomes compared with prior studies. Thirty-two patients underwent PCD plus antibiotics, and 3 patients (9.4%) eventually died. In a meta-analysis of management strategies, the mortality rate with medical treatment alone was 50%, and for PCD plus medical treatment the mortality rate was 13.5% . The reported mortality rate with PCD is consistent with our results. However, in our study, 8 patients experienced recurrent EPN, and 4 patients had persistent fever after PCD. The failure rate of PCD was 37.5%.
We then investigated the factors that were predictive of mortality and failure of conservative treatment. To the best of our knowledge, this is the first study to specifically analyze the predictors associated with conservative management failure of EPN and offer the recommendation of appropriate empirical antibiotic. It is also one of the largest studies investigating the prognostic factors of EPN.
Diabetes mellitus appeared to be the most common risk factor for developing EPN. It is believed that a high tissue glucose concentration may provide a favorable environment for the growth of gas-producing bacteria and for the inhibition of leukocyte function, which then impairs the response to infection . Nevertheless, diabetes was not found to be associated with increased mortality (p = 0.65, odds ratio [OR] = 0.67) or with failure of conservative treatment (p = 1.00, OR = 1.32) in the present study, even among the patients with poorly controlled diabetes (HbA1C > 8%). Similarly, there were no significant associations between a higher mortality rate in the patients with EPN and sex, age, or other comorbidities such as liver cirrhosis, hypertension, and history of cerebrovascular accidents.
Serum albumin is the most abundant protein in human blood plasma and constitutes approximately half of the blood serum proteins. Albumin transports hormones, fatty acids, and other compounds, buffers blood pH, and maintains osmotic pressure. Albumin synthesis is suppressed in response to inflammatory conditions and malnutrition . In sepsis, infection, trauma, or major surgery, albumin level decreases by 1–1.5 g/dl over 3–7 days. In a study of over 15,000 severely ill patients, an admission serum albumin < 3.4 g/dl has been correlated with increased of mortality . Hypoalbuminemia was found to be a strong predictor of increased mortality and morbidity in surgical or ICU patients . It has been reported that survival after acute kidney injury is significantly altered according to the level of serum albumin . Accelerated protein breakdown is a feature of the metabolic alterations seen in patients with acute kidney injury. In our previous study, severe hypoalbuminemia (<3.0 g/dL) combined with acute kidney injury was associated with an increased mortality rate (42.9%, p = 0.03) in patients with EPN. In the present study, severe hypoalbuminemia was independently associated with an increase in the failure of conservative treatment. Even in the patients treated with PCD plus antibiotics, the failure rate was still independently associated with severe hypoalbuminemia. Therefore, the patients with severe hypoalbuminemia had a higher risk of conservative treatment failure and additional management including salvage nephrectomy, open drainage or repeated PCD was required. Hypoalbuminemia is probably related to infection, however, it can also be the result of a previous illness, such as liver cirrhosis. When the patients with liver cirrhosis was excluded, there was still association between PCD failure and severe hypoalbuminemia (n = 30, p = 0.01). It is not known whether albumin concentration itself is important or whether it is simply a marker of poor general health. The benefit of correcting hypoalbuminemia in this situation has not been clearly established. However, PCD with antibiotics is still recommended in the initial stage for the patients who have severe hypoalbuminemia, as it immediately reduces renal tissue pressure and decreases the mortality rate.
In a literature review, many prognostic factors for mortality were identified, however none of the trials studied a large population. Khaira et al.  also reported that shock was an independent poor prognostic risk factor in a case series of 19 patients with EPN. Huang and Tseng  reported that thrombocytopenia, altered mental status, severe proteinuria, and acute renal failure at the presentation of EPN were associated with a poor outcome. Similarly, a study conducted in India of 39 patients with EPN showed that altered mental status, thrombocytopenia, renal failure, and severe hyponatremia at presentation were also associated with higher mortality rates . In a meta-analysis, systolic blood pressure less than 90 mmHg, serum creatinine greater than 2.5 mg/dL, and impairment of consciousness were also found to be associated with increased mortality . In the present study of 44 patients with EPN, need for emergency hemodialysis, shock on initial presentation, altered mental status, severe hypoalbuminemia, inappropriate empirical antibiotic regimen and polymicrobial infections were significantly associated with mortality. With regards to the classification of EPN based on the degree of gas seen on CT, the outcomes of the patients with class 3 and 4 EPN were not statistically different from those with class 1 and 2 disease, which is in agreement with previously reported studies . Even though Huang and Tseng reported a tendency towards higher mortality and failure rates with PCD in extensive disease (class 3 and 4) , we did not find this association. The alternative approach of conservative treatment also had no association with the failure rate (p = 0.23).
Microbial organisms cultured from the urine, kidneys and blood of patients with EPN include the common urinary pathogens, E. coli, K. pneumoniae and P. mirabilis. In the present study, other organisms cultured consist of P. aeruginosa and Enterococcus species. Infection by anaerobic organisms (for example Bacteroides fragilis and Clostridium septicum) has also been reported [18, 19]. It is common wisdom that appropriate empirical antibiotic treatment (i.e., matching the in vitro susceptibilities of the isolated pathogens) reduces mortality. In EPN patient, we also found that inadequate empirical antibiotic use was significantly associated with mortality. The choice of empirical antibiotic regimen should be appropriate for targeting Gram negative bacteria. Third-generation cephalosporins, for example ceftazidime, is recommended as the initial antibiotic regimen because of high susceptibility and anti-pseudomonal activity. In patients with severe sepsis and septic shock, vancomycin plus imipenem or ceftazidime as the empirical choice is suggested. However, fluoroquinolone should be avoided because of lower susceptibility rate among common gram-negative bacteria in Taiwan . Approximately 70% of the E. coli isolates and 50% K. pneumoniae isolates were susceptible to fluoroquinolones.
This study has limitations inherent to all retrospective, single institution studies. The small number of patients may account for the lack of significance of some of the factors analyzed. Second, we mostly used data acquired at the initial presentation to identify the risk factors for mortality. Factors that varied over the time of treatment may also have influenced outcomes. A larger prospective cohort study is required to support our findings.