The clinical and imaging presentation of acute "non complicated" pyelonephritis: A new profile for an ancient disease

Acute pyelonephritis (APN) is an extensively described, well-known disease. The first descriptions date to ancient Egypt, underlining its severity and its potential to lead to sepsis, kidney abscesses and destruction of the kidney parenchyma [1]. In spite of this long history, the nomenclature of acute pyelonephritis is still controversial and the semantic ambiguities can still cause confusion.

A general consensus is reached only for the definition of complicated versus non-complicated pyelonephritis: the former term refers to the presence of systemic (such as HIV positivity, diabetes, neoplasia or collagen disease) or anatomical (such as active stone disease, obstruction, reflux nephropathy) predisposing factors, the latter term to their absence [1–4].

As Talner and co-workers underlined over 15 years ago, there is no agreement on the terminology of acute pyelonephritis and, as a consequence, on diagnostic definitions [5]. Indeed there are two distinct and legitimate definitions, involving different diagnostic approaches. According to the "classical" one, reported in almost all textbooks of medicine, the definition is pathological, requiring the demonstration by imaging techniques or renal biopsy involvement (only rarely performed, limited to kidney transplants) of the renal parenchyma [5, 6]. According to this definition, upper urinary tract infections (UTI) can be classified as pyelonephritis (with evidence of parenchymal involvement) or pyelitis (without evidence of parenchymal involvement at imaging). Only the pathological approach, based on imaging techniques, is apt to identify the presence of kidney scars. This approach is often followed in children, as the pediatric literature usually underlines the importance of renal scars as a potential cause of progressive kidney damage and hypertension or of pre-eclampsia later in life [6–11]. Nevertheless the issue is controversial and long-term effects of kidney scars have yet to be fully demonstrated [12–14].

The second definition is clinical only. Under the pressure of cost constraints, most Authors in the last decade have preferred (at least in adults) a clinical definition based on a classic tetrad of high fever, costovertebral angle tenderness, signs or symptoms of lower urinary tract infection and positive urinary cultures. This approach has the advantage of simplicity and lower costs but is unable to discriminate between the different forms of upper urinary tract infections (upper UTI) [6, 15–17].

The therapy of upper urinary tract infections is likewise controversial, possibly because of the different prevalence of severe parenchymal lesions in the different series. General agreement is reached only on the potential danger of kidney abscesses, requiring hospitalization and long-term antibiotic treatment [6, 15–17].

In this context, there are at least two good reasons to reconsider acute pyelonephritis one century after the pivotal description by Tiemlich and a decade after the K-DOQI re-definition of chronic kidney disease (CKD) [1, 18].

The first is the re-definition by the K-DOQI guidelines of CKD as any persistent alteration of the morphology of the kidneys or of kidney function. Thus kidney scars define the presence of CKD, suggesting that the therapeutic goals should include their avoidance, along with the prevention of sepsis and renal abscesses [18].

The second reason regards the impressive improvements of imaging in kidney infections, allowing the identification of small lesions and the distinction between healing, fibrotic and edematous (presumably active) infectious foci [19–30]. However, despite the improvement in diagnostic tools, very few recent studies have investigated the relationship between clinical and imaging presentation in adults. Our previous studies suggested a relationship between the presence of renal abscesses at diagnosis and the development of kidney scars. They also indicated, in a cohort of cases referred to the Emergency Room, that only imaging data could distinguish between upper urinary tract infections with and without parenchymal involvement [31–33].

The aim of the present prospective study (June 2005-December 2009) was to analyze the clinical and imaging presentation of all patients hospitalized in our setting, in which diagnosis of APN was made according to imaging criteria.

Our goal was to test the relationship between the clinical-laboratory data at presentation and the imaging data. We reasoned that a strong correlation would allow a simpler, more rapid identification of the cases with the severest kidney involvement (abscessed and/or multiple lesions), requiring hospitalization and long-term therapy, while the lack of correlation would support the wide use of imaging techniques.

The present study was performed in a setting in which patients with a clinical suspicion of "non-complicated" APN are routinely hospitalized and undergo an extensive imaging work-up (ultrasound and either MR or CT scans) [31–33].

In two recent surveys in the USA and Korea, the incidence of APN was reported as 20-35 new cases per year per 10,000 inhabitants [34–37]. Thus the gross incidence was about 10 times higher than in our study (about 2.5 new cases per year per 10,000 inhabitants). However it is difficult to compare the incidence data in our study with those in the literature, for two main reasons. Firstly, the definition of APN rests on different bases. Imaging data are not included in the diagnostic work-up in the USA and Korea but are required for diagnosis in our setting. Secondly, in the absence of an imaging work-up, the incidence data in the USA and Korea include both non-complicated and complicated pyelonephritis while our study was limited to the former category. In our setting the ratio between "complicated or secondary" and "non-complicated or primary" APN is approximately 3:1 in the Emergency Room (ER), a finding that could at least partly explain the difference.

Unlike other settings, Italian ERs are usually the second referral point after the family physician; both family physicians and ERs are provided by the national health care system and are free of charge. An efficient filter by family physicians could be postulated on the basis of the 42% prevalence of antibiotic treatment in the last 3 days and of a median time of 3 days between the onset of symptoms and hospitalization.

In this context the present study was planned with the aim of describing the clinical presentation of imaging-confirmed APN. From the practical point of view we wanted to identify clinical or biochemical markers of severity of the kidney lesions (multifocal and abscessed lesions) to test the alternatives to our imaging-guided diagnostic and therapeutic approach.

The key feature of our study is the high prevalence of severe lesions: multiple lesions were present in 79.8% of the patients and abscessed lesions in 39.5%.

In spite of the severity of the lesions we found a high prevalence of incomplete clinical presentations, lacking the full-blown syndrome classically described as high fever, signs or history of recent lower UTI, costovertebral flank pain or tenderness and positive urinary cultures (table 2). The three symptoms were present at hospitalization in about one third of cases and lower urinary tract symptoms were present in about half of the patients (47%). These clinical hallmarks are usually included in the enrollment criteria for clinical trials for APN, together with positive urinary cultures and no recent antibiotic therapy [6]. According to these criteria, only about 10% of our patients would have been eligible for a recent clinical trial on APN (table 2).

The frequent negativity of the urinary cultures deserves further comment. About half of the cases were treated with antibiotics but the difference between cases with positive and negative cultures was only marginally significant (p = 0.051). The relatively long interval between onset of symptoms and hospitalization and the frequent lack of lower urinary tract symptoms suggest alternative and adjunctive explanations.

Interestingly, a similar prevalence was found in a previous study, performed in a different setting (the largest University Hospital of the Region) in 2004-2006, in which the prevalence of positive urinary cultures was even lower (20.7%), albeit with a higher recall of antibiotic therapy (95%) [32]. This observation may indicate a "referral" bias (patients are usually referred to the emergency room after a few days of empirical therapy), more than a laboratory bias.

To further assess this point, we planned an analysis in the Emergency room to test the hypothesis that, in some cases, cultures may have been performed after the first dose of intravenous antibiotics (as may occur overnight or during the weekend, when urine sampling may be erroneously postponed in a crowded ER). The analysis is still ongoing; however even if the sampling may have been biased in some cases, this finding draws attention to the fact that urinary cultures can become negative after a single antibiotic dose, even in the presence of renal abscesses or multiple lesions.

It is difficult to explain the higher prevalence of positive cultures in non-abscessed lesions. Our hypothesis is that this may be due to the fact that the bacteria are "sequestered" in abscessed lesions. However further analysis on larger cohorts is needed to confirm this apparent paradox.

The difficulty in diagnosing APN is further underlined by the fact that over 15% of the patients displayed renal scars at the first imaging test, presumably as a result of previous APN episodes; APN had been previously diagnosed in only a minority of them (3/18).

In this context and within these limits, the major finding of our study is the lack of correlation between the severity of the renal lesions and the clinical and biochemical data at presentation (table 1, table 2, table 3 table 4). In fact oligosymptomatic presentation may coexist with abscessed or multifocal lesions and none of the tested parameters discriminated the severity of parenchymal involvement. This is of particular relevance for abscessed lesions which require long-term antibiotic therapy and are at high risk for the development of kidney scars [15–17, 23, 25, 32].

Like all clinical studies, ours has strengths and weaknesses. The main strengths are the novelty and the systematic referral pattern with a diagnostic use of second-line imaging techniques. To our knowledge, this is the largest series of adults published in the last 10 years, combining a clinical and imaging work-up of all cases of suspected APN referred to the ER of a relatively large hospital in a short time span.

The limitations are partly shared by single-center observational studies. In the absence of a control group and of epidemiological data on the world-wide incidence of "non-complicated" APN, our data are hardly comparable with the literature. The suggestion that we may be dealing with negative selection of the cases, possibly due to the extensive filter of family physicians, needs further confirmation on a larger scale and in different settings. A potential confounder is the high prevalence of negative urinary cultures, possibly reflecting the policy of empirical antibiotic treatment before hospital referral and the lack of lower urinary symptoms but also perhaps related to late testing. This issue needs further analysis, also in view of the lack of recent epidemiological data on the prevalence of negative urinary cultures in APN [15–17, 33–38].

These limitations may become suggestions for further studies on the prevalence and type of parenchymal involvement and on the clinical presentation in settings with different incidence and referral patterns.

The diagnosis of APN is still a challenge. At least in settings such as ours in which patients are referred to the Emergency Room after a few days of empirical treatment, the prevalence of incomplete oligosymptomatic presentations is high.

In a context of high prevalence of abscessed lesions (39.5%), severe kidney involvement coexisted with incomplete presentations. None of the clinical or biochemical markers at presentation allowed discrimination between small simple lesions and multifocal or abscessed ones. Hence imaging techniques were needed to assess the severity of kidney involvement and to plan the antibiotic therapy. The relatively high prevalence of kidney scars at presentation (15.1%), mostly with a silent clinical history, draws attention to the difficulty in diagnosing and preventing long-term sequelae of APN.

Therefore, we believe that two recommendations for clinical practice can be drawn from our experience. First, the differential diagnosis of APN should be considered in all patients presenting at least one suggestive clinical feature (fever, flank pain, lower UTI symptoms) accompanied by high inflammatory markers. Second, second-line imaging tests (MRI or CT scans) should be systematically used to define the prevalence and type of parenchymal lesions in order to tailor interventions to the specific clinical contexts.