Study setting and main diagnostic pathway
All consecutive patients hospitalized for acute pyelonephritis in the Urology and Nephrology, Internal Medicine and Emergency Medicine wards of the hospital san Luigi Gonzaga in the period June 2005-December 2009 were included in the study. All patients were referred to the Nephrology Unit during hospitalization for the planning of subsequent follow-up and were followed in the Nephrology Unit at least until clinical and radiological healing.
Data were gathered prospectively, starting from the first clinical visit to the Nephrology Unit.
The san Luigi Gonzaga Hospital is a 300-bed university hospital (350, including day-hospital beds); the geographic area covers approximately 100,000 inhabitants. All patients were hospitalized from the hospital Emergency Room (ER); the ER performed over 41,000 visits per year in the first period (May 2005-May 2006) and over 55,000 per year in the last one (January 2009-December 2009).
Clinical definitions
The diagnosis of APN was made by the ER physicians; diagnosis was based on the presence of at least one of the three main symptoms (fever, costovertebral angle tenderness or pain, recent or present UTI) and at least one sign of systemic infection (high WBC or C-reactive protein). The presence of leukocytes or nitrites was an ancillary criterion supporting the presence of UTI. The presence of shivers, routinely evaluated in the Emergency Room as part of the patients' work-up, was not uniformly recorded and was not considered as sensitive enough to be added to the selection criteria.
The diagnosis of present or recent UTI was mainly based on the clinical history and on the clinical symptoms; in fact, in our setting, patients usually start an empirical therapy (either prescribed by the family physician or self-prescribed) before undergoing urinary tests, in particular if they are prone to develop UTI. This is also the reason why, in our opinion, the prevalence of positive urinary cultures was relatively low.
All patients with a clinical picture suggestive of APN underwent abdominal ultrasound, whenever possible in the ER, and were hospitalized. In patients hospitalized for fever of unknown origin, in which the diagnosis was made during hospitalization, ultrasound was performed subsequently.
In the presence of systemic predisposing factors (any factor affecting the immune response, including diabetes, collagen diseases, chemotherapy, HIV positivity, neuromuscular diseases) and/or of anatomical predisposing factors (any factors causing obstruction, including active stone disease, prostatic hypertrophy, kidney malformations, polycystic kidney disease and indwelling catheters) the APN was considered as "complicated" or secondary and the patients were not included in the present analysis. Regarding the diagnostic pathway, in the presence of anatomical predisposing factors the most widely employed technique was CT scan, better in defining the urinary tract and superior in the definition of stone disease. Patients with complicated pyelonephritis were followed by the urological team and were not included in the analysis. Patients with systemic predisposing factors (such as diabetes or neoplasia) were followed by our group but were not included in the analysis, as the predisposing factors could affect both presentation and follow-up. Pregnant patients were also excluded. History of kidney stones was recorded, and APN was considered as non-complicated in the absence of kidney stones at imaging at diagnosis or of a history of having passed a kidney stone in the previous month. The only morphological exception was the presence of kidney scars presumably due to previous APN episodes, on the basis of the spatial relationship with the excretory system, at the first imaging.
Consent for publication and for anonymous management of data was obtained either at hospitalization or at first access in the day-hospital. Prof. Alberto Angeli from the Ethical Committee of the University of san Luigi Gonzaga assessed the study and confirmed that ethical approval was not needed since the study deals with a non-invasive diagnostic tool used according to its standard indications in a specific subset of patients, for which previous indications of the literature are available. Furthermore the study analyzes the results obtained in clinical practice and no test was performed for the sake of the study. Approval for the indication of each specific imaging test is a part of the routine controls of the Magnetic Resonance or Computerized Tomography experts, who ask for informed consent for each test and, in each case, control for the coherence of prescription of the analysis.
Imaging criteria
The diagnosis of pyelonephritis required demonstration of involvement of the kidney parenchyma. In cases with clinical suspicion of APN, in which no predisposing condition was found at ultrasounds, a "second-line" imaging test was performed. Since patients with non-complicated APN are mostly women of childbearing age, MR was chosen as the preferred imaging technique and CT scan or scintigraphy was performed only in the case of contraindications or logistical problems (long wait before availability of MR scan). As a rule, imaging data were obtained within 3 days after hospitalization.
In view of the low sensitivity of ultrasounds for the presence of parenchymal lesions and the high sensitivity for obstructive lesions, the ultrasound imaging was used to identify the presence of anatomical predisposing factors.
At the MR scan, the diagnosis of APN requires the demonstration of the following: in the basal imaging test, the presence of minor alteration with non-homogeneous hypointensity in T1-weighted sequences and hyperintensity in T2-weighted sequences. After the injection of contrast medium (Gadolinium), in T1-weighted sequences, reduction of parenchymal contrast enhancement in the affected area is observed. Abscessed areas are defined by a peripheral halo enhancement after contrast medium.
Diffusion (DWI) sequences were also studied. APN is characterized by areas of reduction of the diffusivity of the water molecules (hyperintense in sequences b = 600); however the finding is less specific [28].
At the CT scan, the diagnosis of APN requires the demonstration of the following alterations after contrast medium: decreased enhancement with a "striped" effect of the renal parenchyma; the triangular shape originally described (lobar nephronia) in the area of decreased perfusion is typical, but not required for diagnosis.
Kidney abscesses are defined by the presence of a hyperintense peripheral border, with a marked hypodensity of the lesion (necrosis), after contrast medium.
Ancillary signs of APN, both at CT and MR scans, are: kidney swelling, perinephric inflammatory fluid and stranding of the perinephric fat and occasionally a minimal dilation of the urinary tract [28].
At DMSA-scintigraphy, the diagnosis of APN requires the demonstration of areas of altered perfusion; the definition of kidney abscesses is not possible by this technique.
Data collection and analysis
The following general data were collected: age, sex, parity, country of origin, previous UTI, previous history of pyelonephritis, previous stone disease. By definition, patients with kidney stones were not included. At referral, the following clinical data were collected: fever (presence, considered as > 37°C; maximum level on the day of hospitalization), costovertebral angle-tenderness, lower UTI symptoms, other symptoms; antibiotic therapy within the last 72 hours; other therapies.
Laboratory data were obtained by standard laboratory methods. The following data, obtained in the Emergency Room at hospitalization, were considered: serum creatinine, CRP, blood cell counts, urinary culture (available in all cases), hemocultures and results of dipstick urinalysis were also recorded when available.
On the basis of their radiological appearance, the lesions were dichotomized in "simple" versus abscessed (at least one abscessed lesion) and in single versus multiple. Further imaging parameters were unilateral versus bilateral involvement and presence of kidney scars at diagnosis.
Informed consent for publication of the clinical data in anonymous form was obtained from all patients at the first nephrological control. Further specific informed consent was obtained from the patient whose images are depicted in Figures 1 and 2.
Statistical analysis
Descriptive analysis was performed as appropriate (mean and standard deviation for parametric data and median and range for non-parametric data). Student's t-test, chi-square test and ANOVA were used for comparisons between groups. Significance was set at < 0.05.
Univariate odds ratios and multivariate regressions were calculated in SPSS (version 17.0). Logistic regression was performed considering the type of lesion (abscessed lesion) or the number of lesions (multiple lesions) as outcomes, with respect to the main clinical data (age, fever, costovertebral tenderness, lower urinary tract symptoms, duration of symptoms and their combination), presence of renal scars and C-reactive protein as a marker of inflammation. Continuous data were dichotomized at the median.