A scoping review of fever of unknown origin with normal serum C-reactive protein

Introduction

Fever of unknown origin (FUO) are still among the most challenging of conditions to diagnose [1], 2]. Petersdorf and Beeson (1961) defined FUO as being 38.3 °C or higher, lasting more than three weeks, and remaining undiagnosed after one week of hospital tests [3]. In 1991, due to the expansion of the range of tests available to outpatients, Durack and Street revised the last criterion to three days of hospital tests or three outpatient visits [4]. Furthermore, in 2009 Vanderschueren et al. defined “inflammation of unknown origin (IUO)” as a condition involving an unexplained increase in inflammatory markers of more than three weeks’ duration that does not meet the definition of FUO [5]. They suggested that FUO and IUO have a similar differential diagnosis, including the ‘big three’ categories of infection, malignancy, and non-infectious inflammatory disorder, thus enabling the same diagnostic strategy to be applied [5]. However, the differential diagnosis and appropriate diagnostic approach to FUO without inflammation remain unclear. If the differential diagnosis of FUO without inflammation can be made more specific, the condition may be able to be diagnosed with greater accuracy, timeliness, and accountability to the patients.

C-reactive protein (CRP) is one of the acute-phase reactant proteins most widely used as an inflammation marker as its serum level increases during both acute and chronic inflammation [6], 7]. Including CRP in the routine diagnostic workup for FUO is recommended on the grounds of its high sensitivity for infectious and non-infectious causes of inflammation [1], [8], [9], [10]. Serum CRP is indeed elevated in most FUO cases whether these are caused by an infectious or rheumatic disease or a malignancy [11], [12], [13], [14], [15], [16], [17], [18], [19]. However, the utility of CRP in the differential diagnosis of FUO remains largely unknown [10]. Among patients with FUO with elevated CRP, the degree of increase in the CRP value has little utility for the differential diagnosis [12], 13]. However, we hypothesized that a completely normal serum CRP value might have a specific diagnostic yield because it usually indicates the absence of a significant inflammatory process. Furthermore, FUO without increased CRP may be more likely to remain undiagnosed than FUO with increased CRP [20], suggesting that the former may require a different diagnostic approach. For example, FUO with normal CRP limits the diagnostic yield of ¹⁸F-fluorodeoxyglucose positron-emission tomography with computed tomography (PET-CT) [21], [22], [23]. Therefore, epidemiological data are essential to improving the diagnosis of FUO with normal CRP.

A scoping review is more suitable than a systematic review if the aim is to identify knowledge gaps, scope a body of literature, and clarify concepts [24]. A scoping review is also more appropriate than a narrative review if clarification of a concept or theory is required [24]. For these reasons, a scoping review is ideal as means of inventorying studies of FUO with normal CRP and identifying any gaps in the current knowledge of the topic [25]. The present review aimed to facilitate future studies by proposing the concept of “FUO-NIL” (FUO with No Inflammatory Laboratory evidence).

Methods

The primary question that guided this review was, “What is the differential diagnosis of FUO-NIL in adult patients?” The secondary questions included, “What is the distribution of the disease categories in FUO-NIL?”, “What tests should be performed in the diagnostic workup for FUO-NIL?”, and “What are the patient characteristics associated with the condition?”

This review, which was conducted in accordance with the Joanna Briggs Institute (JBI) methodology for scoping reviews [26] and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Extension for Scoping Reviews (PRISMA-ScR) guidelines [27], identified case reports and other studies of FUO with normal serum CRP. The requirement for ethics approval was waived because no human participants were recruited. The review protocol has been submitted to the Open Science Framework database for publication (www.osf.io/v5qbj).

Results

After duplicates were removed, a total of 3,084 citations were identified from searches of electronic databases and review article references (Figure 1). Based on the title and the abstract, 2,741 were excluded, with 342 full text articles to be retrieved and assessed for eligibility. Of these, 325 were excluded for the following reasons: 6 were not studies about FUO, 5 did not include adult patients, and 314 included only patients with positive or unknown CRP levels. We excluded 1 study because we were unable to retrieve it [28]. Ultimately, 17 articles met the inclusion criteria after a full-text review [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45]. The discordance rate between the two reviewers was 4.5 % (140 of 3,084 items) during the initial screening using titles and abstracts and 3.2 % (11 of 342 items) during the secondary screening based on the full-text review. Of the 17 studies included, 13 were case reports [29], [30], [31], [32], [33, 35], [39], [40], [41], [42], [43], [44], [45], three were case series [36], [37], [38], a and one was a cross-sectional study [34]. Four of the case reports were conference abstracts [29], 35], 39], 42]. In one case series, the specifics of a CRP-negative case not described in the text were ascertained by contacting the authors via email [38]. All but one article, which was written in Korean, had been published in English [31].

Figure 1:

Flow chart of study inclusion.

A cross-sectional study of 41 cases of functional hyperthermia was excluded from the analysis because the inclusion criteria were unclear in each case, making it difficult to determine how many cases satisfied the inclusion criteria of the present study [34]. A case report written in Korean with a final diagnosis of “Janggam disease” [31] was also excluded because the disease is not recognized in Western medicine. In total, 19 cases of FUO-NIL were finally identified for analysis.

Table 1 summarizes the demographic variables and clinical data. Six patients had a history of human immunodeficiency virus infection [36], 43], one patient had a history of abatacept use [35], and one patient with an allogeneic stem cell transplantation had a presumptive history of immunosuppressive therapy despite the absence of any mention of this in the report [39]. PET-CT findings for four patients were documented [29], 32], 41], 43]; all four were positive, with the findings being critical for the diagnosis in three patients. Only two reports explicitly stated that the CRP value remained negative across multiple measurements [39], 44]. In other reports, either the number of CRP measurements was unclear or only the CRP value at the initial presentation was documented. In all the cases except five autopsy cases in a single study in which the timing of CRP measurement was unclear, CRP was measured at least once before the initiation of disease-specific treatments. The white blood cell (WBC) and erythrocyte sedimentation rate (ESR) values were specified for only six and three patients, respectively, with none of the patients having an elevated WBC count. The five autopsy cases in the study of disseminated toxoplasmosis in AIDS were counted as deaths [36]. The total proportion of surviving patients was 73.7 %.

Table 2 lists the final diagnosis of the 19 cases. As mentioned above, the cross-sectional study of 41 cases of functional hyperthermia was excluded from the analysis but retained on the list as offering a potential differential diagnosis of FUO-NIL. The diagnostic categories were infection (n=10, 52.6 %), malignancy (n=3, 15.8 %), non-infectious inflammatory disease (n=4, 21.1 %), and some other cause (n=2, 10.5 %). Four patients had hemophagocytic lymphohistiocytosis (HLH) or macrophage activation syndrome (MAS) [35], 40], 42], 43]; however, three of these patients clearly had tuberculosis [40], systemic lupus erythematosus (SLE) [42] or lymphoma [43] as the primary disease, which were therefore treated as the final diagnosis.

Discussion

Understanding atypical presentations of diseases and syndromes is crucial for diagnostic reasoning. While FUO often involves inflammation, atypical cases without laboratory evidence of inflammation deserve close attention. The present scoping review systematically explored the novel idea of “FUO-NIL,” defined as FUO with normal serum CRP, which as of yet lacks an established diagnosis, management method, and description of outcomes. This review will hopefully create a foundation or platform for advancing the current understanding of, and guide future research into, FUO-NIL.

The findings of the present review demonstrated that FUO-NIL is a heterogeneous condition with diverse, underlying etiologies. The disease categories among the 19 cases which were eligible for quantitative analysis were infection (n=10, 52.6 %), malignancy (n=3, 15.8 %), and non-infectious inflammatory disease (n=4, 21.1 %), which together accounted for 89.5 % of the cases. These proportions were largely consistent with those reported in previous studies of FUO [9]. However, no certain conclusion can be drawn owing to potential biases, including the publication bias. Most of the diagnoses shown in Table 2, including tuberculosis, intracellular bacterial infections, such as brucellosis and Q fever, SLE, and lymphoma, are ‘classic’ causes of FUO, thus corroborating the recommendation of previous cohort studies of FUO that a high index of suspicion be maintained for these diseases [9], 46].

In terms of infection category, common, infectious causes of FUO, such as infective endocarditis, prostatitis, and abscesses [8], were notably not ascribable to FUO-NIL. Given previous reports of the limited sensitivity of CRP in distinguishing infectious from non-infectious diseases [47], this result was somewhat unexpected, but the apparent increase in the sensitivity of CRP may be explained by the definition of CRP negativity (<0.5 mg/dL) employed, which was stricter than that used in earlier studies [48], 49]. The majority of infectious causes comprised intracellular, bacterial, zoonotic infections, including brucellosis, Q fever, and toxoplasmosis, and toxocariasis. Tuberculosis, which is usually the most common, infectious cause of FUO [8], 46], was identified in only one patient possibly because intracellular bacteria, which replicate within the host cells, may escape an immune response unlike common pathogens in extracellular bacterial infections, thereby potentially inducing a different pattern of CRP production [50], [51], [52].

In terms of malignancies, both hematological malignancies and solid tumors were found to be potential causes of FUO-NIL. Among the patients diagnosed with malignancy in this study, PET-CT was performed in three cases, and its findings identified biopsy-positive sites in two patients [29], 32]. It is worth noting that one of these two patients had a normal ESR [29], indicating the utility of PET-CT for diagnosing malignancies in patients with FUO even in the absence of inflammatory markers, despite the suggestion in previous studies that PET-CT may not be useful for diagnosis if both CRP and the ESR are normal [22].

Non-infectious inflammatory diseases, such as SLE and granulomatous diseases (sarcoidosis and granulomatous interstitial nephritis), were also identified as causes of FUO-NIL [30], 42], 44], 45]. SLE is widely recognized as an “exceptional” rheumatic disease in that the CRP value is unreliable as a biomarker of disease activity [53]. This observation accords with the notion that SLE may be a representative cause of FUO-NIL. The tendency for elevated serum CRP to be absent in SLE may be attributable to treatment modifications, including the use of corticosteroids, the involvement of immune complex formation containing CRP or other factors [54]. However, both cases of SLE in this scoping review [42], 45] had been newly diagnosed following FUO development, thus rendering treatment-related modification an unlikely reason for the low CRP values. Similarly, the CRP value reportedly correlated weakly with the disease activity of sarcoidosis [55]. While it remains unclear whether this characteristic applies to other granulomatous diseases, the identification of sarcoidosis and granulomatous interstitial nephritis suggested that granulomatous diseases in general should be considered as a potential cause of FUO-NIL.

Fabry disease in the miscellaneous causes category, a condition associated with hypohidrosis, is known to cause fever through thermoregulatory dysfunction by impairing the function of the sweat glands and the cholinergic system [33]. Consequently, diseases which were not identified in this scoping review but are associated with hypohidrosis, such as Sjögren’s syndrome, systemic sclerosis, and acquired idiopathic generalized anhidrosis, also potentially contribute to the incidence of FUO-NIL [56], 57]. HLH is reported to be associated with a CRP level that is significantly lower than that associated with adult-onset Still’s disease [58], 59]. Although the cross-sectional study of 41 cases of functional hyperthermia was excluded from analysis for the previously stated reason, the findings of the cross-sectional study nevertheless indicated the potential significance of functional hyperthermia in the ‘miscellaneous causes’ category of FUO-NIL [34]. However, it should be emphasized that pathophysiologically, hypohidrosis and functional hyperthermia should more accurately be characterized as ‘hyperthermia’ rather than ‘fever’, a distinction which is fundamental to clinical reasoning. Differentiating between these two, distinct conditions, which are currently included in FUO-NIL, awaits future research.

The limited number of articles in this scoping review precluded the development of a comprehensive clinical approach for managing FUO-NIL. However, several, important insights were gained. First, the differential diagnosis of FUO-NIL seems to differ from the typical causes of FUO encountered in clinical practice. While the standard diagnostic workup for FUO should still be performed even when the CRP value is negative, the findings of this review highlighted the need to expand the differential diagnosis. Specifically, potential causes of FUO-NIL should include infectious causes, such as intracellular, bacterial and protozoal infections and tuberculosis; non-infectious, inflammatory diseases, such as SLE and granulomatous diseases; and conditions associated with hypohidrosis, including Fabry disease and HLH, and functional hyperthermia in the category of miscellaneous causes. The clinical evaluation of patients with FUO-NIL should pay special attention to a history of animal exposure and the signs and symptoms of hypohidrosis and psychosocial stressors potentially linked to functional hyperthermia. Furthermore, PET-CT should be considered a part of the diagnostic workup for FUO even if the CRP value is normal. With the exception of one case series concerning disseminated toxoplasmosis, which was diagnosed post-mortem in patients with AIDS, the present review found no fatalities associated with FUO-NIL. However, the inclusion of potentially life-threatening conditions, such as disseminated toxoplasmosis, tuberculosis, malignancies, SLE, and HLH, underscores the importance of an early diagnosis.

This scoping review has several limitations. First, the sample pool was relatively small, reflecting the rarity of FUO-NIL and the limited availability of detailed reports that include the CRP value of the patients. The small sample size limited the generalizability of the findings, emphasizing the need for a larger-scale study. Second, because the inclusion criteria pertained neither to the method nor the device used for measuring serum CRP, the definition of normal CRP employed in this study may not be generalizable. Third, although a normal CRP value was required for inclusion, it was impossible in practice to confirm whether the value remained normal during the entire course of each patient. Moreover, abatacept therapy in a case of rheumatoid arthritis [35] and presumptive immunosuppressive therapy in a case of bone marrow transplantation [39] suggested that these treatments may have influenced the CRP value. Hence, the impact of these treatments was unable to be ruled out. However, by excluding cases of fever with a duration < 3 weeks, the possibility of misclassifying early inflammatory processes as FUO-NIL was likely minimized. Fourth, this review did not directly compare FUO-NIL with inflammatory FUO, thereby limiting our ability to delineate the distinguishing features of the former. Future research should aim to compare these conditions to clarify their differences with a view to refining the diagnostic strategy. Fifth, the findings may have involved a publication bias, as most of the studies included were case reports or case series, which often selectively report rare and diagnostically challenging cases, thereby possibly skewing the distribution of the diseases associated with FUO-NIL. Nevertheless, the primary objective of this review was to define FUO-NIL as a distinct clinical entity and provide a foundation or platform for future research. Sixth, many of the studies included in this review lacked details of the diagnostic tests performed. This omission limited our ability to propose a robust, evidence-based approach to FUO-NIL. The data on the diagnostic tests shown in Table 1 only reflect the tests mentioned in the articles consulted for the present review, which was conceptual in nature and aimed to propose a framework for understanding FUO-NIL rather than offer a definitive conclusion about its diagnosis or management.

Despite these limitations, this scoping review represents an important first step in recognizing FUO-NIL as a distinct clinical entity. It is hoped that identifying the gaps in the current knowledge of this condition will provide impetus to further research into this challenging condition, including the construction of a diagnostic algorithm beginning with the presence or absence of inflammation. To validate the concept of FUO-NIL and to refine its clinical implications, further, large-scale, prospective studies are needed. For example, a prospective cohort study comparing the final diagnoses between a FUO-NIL group, as defined by the absence of the inflammatory markers analyzed in the present study using pre-determined, continuous CRP measurements and a CRP-positive FUO group may be conducted. Such a study may apply a stricter definition of FUO-NIL, such as one that stipulates at least two, consecutive, negative CRP results (<0.5 mg/dL) obtained at an interval of more than one week to increase specificity. The present review did not exclude individuals receiving immunosuppressive or immunomodulatory drugs, who are more likely to have a negative CRP result; future studies should exclude such individuals to reduce bias.

Another topic for debate is the optimal combination of inflammatory markers for defining FUO-NIL. For example, ESR is commonly included in the initial workup for FUO [1], 8], 46] but may be inferior to CRP for assessing the degree of inflammation because the latter more accurately reflects the presence and degree of inflammation while the former can be affected by numerous factors other than inflammation, such as anemia, kidney function, hypercholesterolemia, pregnancy, sex, and advanced age [60]. This is one of the reasons why CRP was chosen in the present review to define non-inflammatory status although there is no warrant that this is the best biomarker for this purpose. Therefore, determining which inflammatory marker or combination of markers is best for defining FUO-NIL should be discussed in future studies.

Conclusions

FUO-NIL may be a distinct, clinical entity that differs from FUO with increased CRP in etiology and diagnostic approach. FUO-NIL has a wide spectrum of causes, including intracellular, bacterial and protozoal infections and tuberculosis; malignancies, such as hematological malignancies and solid tumors; non-infectious inflammatory diseases, including SLE and granulomatous diseases, and other, miscellaneous causes, including Fabry disease, HLH, and functional hyperthermia. The findings suggest the need for vigilance in diagnosis and a tailored, diagnostic strategy, including a focus on history of animal contact, hypohidrosis, and psychosocial stressors, as well as the use of PET-CT to assess for malignancies. Further research is needed to enhance the current understanding of FUO-NIL and improve its diagnosis and management.