Studies addressing extremely high CRP values in dogs are scarce [3] and are even rare in humans [18, 28]. To the authors’ knowledge, this is the first study evaluating extreme increases of CRP in a mixed canine patient population with various diseases. In our study, the decision limit for CRP was chosen in accordance with other human studies which often use cut-off values of 100 mg/l to rule in a bacterial infection and apply antibiotics.
Prevalence of CRP > 100 mg/l
To the authors’ knowledge, there are no studies in dogs primarily aimed to investigate the prevalence of extremely high CRP values above 100 mg/l. However, when evaluating data reported previously for a mixed canine patient population in a veterinary university clinic in Japan, 9% dogs had a CRP > 100 that is in accordance with our findings [19]. In humans, reported prevalence for CRP > 100 mg/l is highly variable and ranges between 3 and 30% [19, 23]. The most likely explanation for the highly variable results is the patient population. While a low prevalence of CRP > 100 will be expected in primary health care facilities, a high prevalence is observed in hospitals with a large number of severely diseased patients.
CRP cut-off to differentiate etiologies
Published human [26] and canine [12, 14, 15] reference ranges, including our own, for CRP measured with different assays are almost similar (< 10–20 mg/l) so that theoretically a similar behavior of both, human and canine CRP could be expected. While bacterial infections have been reported in 55% (462/839) [18] to 78% (35/42) [20] of human patients with CRP concentrations > 100 mg/l, only 22% (33/149) of canine patients in our study fell into this category. In contrast to human beings, high CRP values above 100 mg/l were not able to discriminate between different disease etiologies in dogs in our data set. Our data confirm the theory that CRP cannot be used as a differentiation parameter to distinguish between infectious or even bacterial and non-infectious /non-bacterial diseases in dogs. Therefore, high CRP values are no indication for antibiotic treatment. Extremely high CRP values indicate a severe inflammation with acute phase reaction with various etiologies. The proportion of non-infectious diseases seen here (12%) is comparable to a recent study in humans (rheumatologic diseases ~ 8%) [18]. In contrast, the proportion of neoplastic diseases (16%) is higher in our study compared to human medicine, where these patients only account for ~ 5% [18]. A descriptive study on dogs with various diseases included a high proportion of patients with immune-mediated diseases or cancer among dogs with > 100 mg/l CRP which is in accordance with our results [19].
As infectious etiologies were most often (55%) the reason for extremely high CRP values in humans, it is recommended to rule out infections as a first step of the diagnostic workup [18]. This could also be adopted to veterinary medicine, keeping in mind that especially non-infectious inflammations and neoplasia are also common etiologies. The fact that the diagnosis “bacterial infection” in our study was not always based on a positive microbiological finding might be seen as a limitation of the study. To our own experience and also according to recent research, microbiological analysis is helpful and important but not infallible. Negative results occur in patients with definite bacterial disease [29]. In human medicine, these are 25–60% of the samples [30]. It is acknowledged to start antimicrobial therapy in critically ill patients based on the suspicion of a bacterial infection [31]. In cases of negative cultural results, it is up to the clinician’s decision if antibiotics are necessary [31, 32]. On the other hand, positive microbiological results do not always prove an infection. There may be (pre) analytical problems such as contaminations or an overgrowth with bacteria that have not been the original pathogens. Furthermore, it is not always possible to take microbiological samples due to the anatomical region of inflammation or the critical state of the patient. Therefore, we took the conscious decision not only to include cases with (positive) bacteriological examination results in our study, but also to include cytological and histological findings as well as the clinical course of the patient. This is handled similarly in other studies [33].
In human and animal patients with neoplasia and in absence of an additional disease, major inflammatory reactions are also a characteristic of malignant tumors (e.g. mammary carcinomas, lymphatic neoplasms) [10, 34]. This is in accordance with our study, where CRP values > 100 mg/l were only found in association with malignant and not benign neoplasia. Benign neoplasia is seldom described to cause major inflammatory reactions [9] and might be associated with ulceration in these cases [10, 34]. In human medicine, a proportion of about 25–40% of patients with malignant cancer has associated infectious or non-infectious mild to severe inflammation, which may be accompanied by a systemic acute phase reaction [35,36,37]. A similar explanation might be true for canine patients in general [34] and for the marked acute phase reaction seen in cancer patients in our study. Nevertheless, diagnostics concerning a possible secondary bacterial infection are indicated in patients with tumor and high CRP values [8, 20].
Our study clearly showed that also trauma or surgery without evidence of bacterial infection might induce an extreme increase in CRP concentrations > 100 mg/l. It is well known that traumatic events lead to tissue damage and induce an acute phase reaction within the first 24 h that subsides gradually until the point of time of suture removal [11, 38, 39]. The increase of CRP does not only depend on the degree of trauma [16, 39] but may also depend on the CRP value before surgery, the surgeon and analytical variations [29]. Another limitation of our study, however, might be the fact that bacterial infections could not always be excluded especially if surgery was performed in primarily “non-sterile” organs such as the gastrointestinal tract. This may have led to misclassifications of the disease category.
CRP cut-off to differentiate diseased organ systems
Our findings in dogs demonstrated that extreme increases in CRP cannot be attributed to specific organ systems. Nevertheless, it was obvious that some organ systems tend to be more frequently involved in marked inflammations. On the one hand, these are organ systems which are frequently exposed to bacteria (e.g. the gastrointestinal tract [19, 28, 40]), but on the other hand these are also organ systems which are affected by non-infectious, often immune-mediated, diseases (e.g. the musculoskeletal system [41, 42]). CRP analyses of patients with diseases of the reproductive organs are likely underrepresented in our study as they are primarily referred to the clinic for obstetrics. Interestingly, an organ system with frequent contact to microorganisms (respiratory tract), which is known to present with high CRP values [3, 28] in the case of bacterial infection, was underrepresented in our study. A reason for that could be that patients with bacterial pneumonia, seen in our university clinic, partly present with complicated disease progresses (e.g. pyothorax secondary to septic foreign body in the lung) or suffer from multiple diseases (aspiration pneumonia secondary to another disease, pneumonia due to immunosuppression) and were therefore not always classified as solely “respiratory patients”. Given our results, the attempt to define a cut-off value to detect septic inflammation is only promising for organ systems rarely affected by immune mediated diseases and after ruling out malignant neoplasia with high probability as it has been done previously for the respiratory tract [3].
Prognostic value of CRP
A mortality rate of 37% found here, was markedly higher than reported in humans with CRP > 100 mg/l, where the overall mortality rate during hospitalization was ~ 9% [18]. The markedly lower mortality rate in humans can be explained by the fact that it was only assessed until the day of hospital release [18]. Moreover, the fact that the option of euthanasia is given in veterinary medicine, which leads to higher “mortality” for severely ill patients, has to be taken into account. The arguments (clinical condition, prognosis, financial reasons, management problems, emotional/ private constitution of the owner, and recommendation of the veterinarian) that led to euthanasia were not evaluated here, which is a limitation of the study. The influence of more advanced life-sustaining measures in human medicine might also have contributed to the better outcome reported in humans.
According to our data, there is a high mortality rate in canine patients (80%) as it is known for human cancer patients with CRP > 100 mg/l. However, it has to be taken into account that the mortality (10–40% in humans with CRP > 100 mg/l [18]; > 60% with CRP > 500 mg/l [28]) is also highly dependent on the underlying etiology.
In contrast, canine trauma patients with high CRP values have a comparatively good outcome as demonstrated here. Previous human studies did often not consider patients with traumatic disease etiology at all (tissue damage) [18, 28]. Moreover, there was also a relatively low mortality in patients with inflammatory diseases.
In various etiologies, CRP has proven to be a marker of disease activity or severity and therefore indirectly, a prognostic marker in dogs [40, 43, 44].
The absolute CRP values in our study are not statistically different for survivors compared to non-survivors. The same was described for initial CRP values in dogs with primary immune-mediated hemolytic anemia (IMHA) [45], Ehrlichiosis [4] and for a mixed population of critically ill dogs [46, 47]. CRP has been shown to be a marker of moderate diagnostic value reflecting disease severity in parvoviral enteritis [43, 48]. In contrast to our results and the previous veterinary studies, in human patients with severe sepsis, CRP was significantly higher in non-surviving patients [49] – surprisingly even without an overlap of the patient groups on the day of admission. Also data for the novel COVID-19 virus in humans imply a negative prognostic significance of high CRP values [50].
The most likely explanation why these results differ from our results is the study population. It appears to be logical that in a quite homogenous group with one disease etiology, higher CRP values indicate a higher severity of disease and worse prognosis. Furthermore, the diagnosis “systemic inflammatory response syndrome (SIRS)” is already an advanced state of disease, while in our mixed study population, dogs with different diseases and disease states are included. Additionally, it has to be taken into account that our study population was initially “biased” by the cut-off of 100 mg/l CRP, while other studies investigate the whole spectrum of CRP values [49]. In our study, evaluation of prognosis was only made in a subgroup of patients (patients with known prognosis according to the patient data system), potentially influencing the evaluation results as the clinic may lose contact due to different reasons (owner happy/unhappy, animal healthy/dead, practical reasons).
It has been shown that dogs with marked inflammation tend to need hospitalization inducing high costs for treatment [51]. The high hospitalization rate of 90% seen here, can be explained with the fact that the majority of dogs was clinically severely ill. Nevertheless, there was no association of CRP and hospitalization. Similar findings have been reported for dogs with autoimmune hemolytic anemia or pyometra [7, 52]. The lack of impact of hospitalization on a single CRP result > 100 mg/l demonstrates that the absolute CRP value alone is not a definite prognostic marker but rather indicates a severe disease. Five of the seven (71%) non-hospitalized dogs were cancer patients. In these patients, the decision against hospitalization was probably not made because of a mild disease but due to the unfavorable prognosis. All five dogs were euthanized within the next two weeks.
This study confirms that extremely high CRP concentrations are an indicator of severe systemic inflammation with acute-phase reaction but should not be seen as an indicator of bad prognosis with the consequence of euthanasia. Instead, the consequence has to be an early start of more advanced diagnostics and therapy as well as close monitoring. The actual prognosis for the individual patient is highly dependent on the disease etiology.
Our study was limited by its retrospective nature. A definite diagnosis up to one single detailed etiology could not be made in some cases. Patients were evaluated in different states of disease. Several patients were already pretreated. This is a known problem of medical studies investigating naturally diseased patients, especially if conducted in referral hospitals. In our study, especially the pretreatment with anti-inflammatory or antibiotic agents is a limitation which might have lowered CRP concentrations. Nevertheless, it can be assumed that canine patients presented to our clinic still had an active disease process, even if pretreated. It has been shown that antibiotic pretreatment in dogs with > 24 h respiratory symptoms did not significantly affect the CRP concentration as well as pretreatment with glucocorticoids [3, 6]. Moreover, underlying conditions might have an impact on CRP concentrations. It is known that severe liver failure might lower the CRP value, but only little information is available in literature [53]. Statistical evaluation was hampered by the relatively small patient number in some subgroups so that significances might have been missed.