The Acute-Phase Response
A child spikes a fever of 39.5°C, feels sick, stops eating, and sleeps constantly. This happens long before any pathogen has been cleared. A blood test drawn the same day shows C-reactive protein levels 100 to 1,000-fold above baseline. Neither symptom occurs at the site of infection. Both are ordered from a distance, by the liver and the brain. They respond to chemical signals released from a wound or infected tissue days earlier. This is the acute-phase response: a coordinated, whole-body reprogramming of physiology triggered by local inflammation. The acute-phase response answers a logistical problem that local inflammation alone cannot solve. A localized innate response — Inflammation: The Innate Response to Tissue Damage — can wall off and attack a pathogen at the site of entry. But many pathogens spread systemically. The raw materials for fighting infection (clotting factors, complement components, protease inhibitors, iron-sequestering proteins) are made in the liver, not at the wound. The body needs a way to tell distant organs that an infection is underway. It must redirect metabolism accordingly. This means increasing production of defense proteins, withholding iron from bacteria, and raising body temperature to disadvantage pathogens and boost lymphocyte function. It also means suppressing nonessential activities like feeding and reproduction until the crisis passes. The key insight that unifies fever, acute-phase protein synthesis, and "sickness behavior" is simple. All are downstream of the same small set of circulating cytokines. Chiefly involved are interleukin-6 (IL-6), interleukin-1β (IL-1β), and tumor necrosis factor-α (TNF-α). These were first described as endogenous mediators of local inflammation (Cytokines and Chemokines: Signaling Molecules of Immunity). Here, though, they act as systemic hormones. The liver reads this cytokine signal and switches its protein synthesis program within hours. The hypothalamus reads it and resets the body's thermostat. The bone marrow reads it and releases neutrophils. Complement proteins made anew by hepatocytes reinforce the very effector system described in Complement Effector Functions and Regulation. This article traces that signal from cytokine to gene to phenotype. It also reviews the experiments that established causality rather than mere correlation. By the end, you should be able to explain several things. You should understand why fever is an evolved defense rather than a pathological accident. You should know why C-reactive protein is measured in clinics worldwide. You should grasp why chronic acute-phase activation causes anemia and amyloid disease. Finally, you should see how a single cytokine receptor pathway in hepatocytes can produce a coordinated protein output. That output resembles an emergency broadcast system.
- The acute-phase response is a systemic reaction to infection/injury, driven mainly by IL-6, IL-1β, and TNF-α released from the inflamed site.
- The liver reprograms its transcriptional output within hours: "positive" acute-phase proteins (CRP, fibrinogen, serum amyloid A, hepcidin) rise; "negative" ones (albumin, transferrin) fall.
- Fever is centrally mediated: circulating pyrogens induce prostaglandin E2 in the hypothalamus, resetting the thermoregulatory set point upward.
- CRP binds microbial phosphocholine and activates complement via C1q, linking the acute-phase response directly to opsonization and the classical complement pathway.
- Hepcidin-driven iron sequestration ("nutritional immunity") starves pathogens of iron but causes the anemia of chronic disease when the response becomes chronic.
- Classic experiments — cytokine injection studies, IL-6 knockout mice, and the discovery of endogenous pyrogen — established causal, not just correlative, links between cytokines and each acute-phase feature.
A handful of cytokines released at a site of infection (IL-6, IL-1β, TNF-α) act as systemic hormones that reprogram liver protein synthesis, reset the brain's thermostat, and redirect whole-body metabolism toward host defense.
The core idea
The acute-phase response (APR) is the sum of systemic physiological changes that occur within hours to days after infection, tissue injury, trauma, surgery, or certain cancers. It is distinct from — but triggered by — local inflammation. Local inflammation is a tissue-level program: vasodilation, increased vascular permeability, and leukocyte recruitment at the site of damage. The acute-phase response is an organism-level program: it involves organs (liver, hypothalamus, bone marrow, adipose tissue, muscle) that are not themselves infected, but that alter their function because they receive a hormonal signal from the infection site. The central signal is a set of pro-inflammatory cytokines — historically grouped as "endogenous pyrogens" — of which IL-6 is now understood to be the dominant driver of the hepatic component, with IL-1β and TNF-α acting both directly on target tissues and indirectly by inducing IL-6 production from macrophages, endothelial cells, and fibroblasts. These cytokines enter the bloodstream from the inflamed site and act on distant receptors: IL-6 receptors on hepatocytes, IL-1/TNF receptors on hypothalamic vascular endothelium, and cytokine receptors on bone marrow stromal cells and myocytes. The response has three classically described arms: (1) hepatic acute-phase protein synthesis, (2) fever, and (3) metabolic and behavioral changes collectively called sickness behavior (anorexia, somnolence, reduced grooming/activity, and in severe cases cachexia). Acute-phase proteins are conventionally divided by the direction and magnitude of their change in plasma concentration. Positive acute-phase proteins increase; they are further subdivided into major (rising 100- to 1000-fold, e.g., CRP and serum amyloid A in humans), moderate (2- to 5-fold, e.g., fibrinogen, haptoglobin, α1-antitrypsin, ceruloplasmin), and minor responders. Negative acute-phase proteins — albumin, transferrin, transthyretin — fall, in part because hepatocyte transcriptional machinery is reallocated toward positive acute-phase protein mRNAs, and in part because falling levels serve specific purposes (lower transferrin limits iron available to circulating bacteria; lower albumin frees amino acid precursors and lowers oncotic pressure to favor exudation). It is worth stating as a simplification: the magnitude and even the direction of an acute-phase response for a given protein can differ between species (CRP is a dramatic responder in humans and rabbits but only modestly increased in mice, which instead rely heavily on serum amyloid P).
Mechanism
Step 1 — Local cytokine production. At the site of infection, tissue-resident macrophages and mast cells detect pathogen-associated molecular patterns via pattern recognition receptors (Toll-like receptors, NOD-like receptors) and, upon activation of NF-κB and inflammasome pathways, secrete IL-1β, TNF-α, and IL-6. Endothelial cells and fibroblasts at the site amplify this by producing additional IL-6 in response to the initial IL-1β/TNF-α wave.
Step 2 — Systemic distribution. These cytokines diffuse into local blood and lymphatic vessels. Unlike chemokines, which mainly act over short distances to guide leukocyte migration, IL-6 in particular is stable enough and produced in sufficient quantity to reach circulating concentrations that act on organs far from the inflamed site — this is what qualifies it as an endocrine-like signal rather than a purely paracrine one.
Step 3 — Hepatocyte reprogramming. IL-6 binds its receptor complex (IL-6Rα plus the signal-transducing gp130 subunit) on the hepatocyte surface. This triggers JAK1/JAK2-mediated phosphorylation of STAT3, which dimerizes, translocates to the nucleus, and binds STAT3-response elements in the promoters of acute-phase protein genes — CRP, serum amyloid A, fibrinogen, haptoglobin, hepcidin. IL-1β and TNF-α act mainly through NF-κB and act synergistically with IL-6/STAT3 signaling; some genes (fibrinogen, haptoglobin) require both signals for maximal induction, while others (CRP, serum amyloid A) are largely IL-6/STAT3-dependent. Within 4–6 hours, hepatocyte mRNA pools shift dramatically; plasma CRP, which is near-undetectable at baseline (<1 mg/L), can exceed 100–500 mg/L within 24–48 hours — among the fastest and largest dynamic ranges of any human plasma protein.
Step 4 — Fever induction. Circulating IL-1β, TNF-α, and IL-6 act on the vascular endothelium of the organum vasculosum of the lamina terminalis (OVLT), a circumventricular organ in the hypothalamus that lacks a complete blood–brain barrier. This induces endothelial cyclooxygenase-2 (COX-2) to convert arachidonic acid into prostaglandin E2 (PGE2). PGE2 crosses into the hypothalamic preoptic area and binds EP3 receptors on thermoregulatory neurons, raising the hypothalamic "set point" for core body temperature. The body now perceives its normal temperature as too cold: peripheral vasoconstriction, shivering (increasing heat production via muscle activity), and behavioral heat-seeking follow, driving core temperature up to the new set point — typically 38–40°C in humans, occasionally higher.
Step 5 — Iron sequestration (nutritional immunity). IL-6 also induces hepatocyte production of hepcidin, a small peptide hormone that binds the iron exporter ferroportin on macrophages, duodenal enterocytes, and hepatocytes, triggering its internalization and degradation. This traps iron inside cells, lowers serum iron, and denies extracellular bacteria (many of which require iron for growth) an accessible supply — a direct link between the acute-phase response and antimicrobial defense that has real clinical consequences when it becomes chronic (see below).
Step 6 — Bone marrow and systemic metabolic effects. IL-6, IL-1β, and TNF-α act on bone marrow stroma to boost neutrophil release and production (a left shift and, over days, granulopoiesis), and on skeletal muscle and adipose tissue to promote proteolysis and lipolysis, mobilizing amino acids and fatty acids as substrate for hepatic acute-phase protein synthesis and immune cell proliferation. TNF-α and IL-1β acting in the hypothalamus and brainstem also suppress appetite and promote lethargy — the sickness behavior component, which can be reproduced experimentally by direct injection of these cytokines into the brain or periphery.
The evidence
Endogenous pyrogen and the Beeson experiments (1940s–1950s). Before any cytokine was cloned, the phenomenon of fever pointed to a circulating "endogenous pyrogen." Paul Beeson and colleagues showed that peritoneal exudate cells (granulocytes) from rabbits, when incubated in vitro, released a heat-labile substance into the supernatant that, when injected into a second, uninfected rabbit, caused fever within the hour — even though no pathogen or bacterial product was present. This was a crucial experimental separation: fever could be produced purely by a host-derived molecule, not directly by microbial toxin. It established that fever is an actively generated host response, not simply a side effect of infection, decades before IL-1 was purified (Charles Dinarello's group identified and sequenced human IL-1 in the early 1980s, confirming it as one of the major endogenous pyrogens).
Cytokine injection studies establishing causality for the whole triad. Direct injection of purified or recombinant IL-1β, TNF-α, or IL-6 into experimental animals reproduces each arm of the acute-phase response individually, allowing dissection of which cytokine drives which effect. Intravenous IL-6 in humans and animals reliably induces CRP and fibrinogen synthesis without requiring IL-1 or TNF co-administration, whereas IL-1β and TNF-α alone induce fever and sickness behavior efficiently but are comparatively weak direct inducers of hepatic acute-phase genes — their major hepatic effect is largely indirect, via induction of IL-6. This dissociation is the strongest evidence that IL-6 is the dominant, largely non-redundant driver of the hepatic acute-phase protein program, while IL-1β/TNF-α are dominant for fever and behavior (though IL-6 also contributes to fever).
IL-6 knockout mice. Genetic loss-of-function is the cleanest test of necessity. Mice engineered to lack functional IL-6 (Kopf et al., 1994, and subsequent studies) show a severely blunted acute-phase protein response after tissue injury or infection — plasma CRP-analog and fibrinogen induction that is largely absent — while retaining much of the local inflammatory response (leukocyte recruitment, TNF/IL-1 production at the wound). This showed that IL-6 is specifically necessary for the systemic hepatic component of the response, dissociating it from local inflammation itself. IL-6 knockout mice also show impaired, delayed fever responses to some (but not all) inflammatory stimuli, consistent with IL-6 acting as one of several redundant fever mediators alongside IL-1 and TNF.
Antibody and receptor-blockade evidence in humans. Tocilizumab, a monoclonal antibody blocking the IL-6 receptor, is used clinically for rheumatoid arthritis and cytokine release syndrome. Its use provided an unplanned but rigorous human "knockout" experiment: patients on IL-6 receptor blockade show blunted or absent CRP responses even during active infection, sometimes masking the clinical signs clinicians rely on to detect sepsis — a striking real-world confirmation, in humans, that IL-6 signaling is required for the CRP response, and a cautionary tale about interpreting acute-phase markers in immunomodulated patients.
Discovery that CRP activates complement. In 1974, Kaplan and Volanakis showed that CRP, once bound to its ligand (phosphocholine on the surface of Streptococcus pneumoniae and other microbes, or on damaged host cell membranes), could bind C1q and activate the classical complement pathway in a manner functionally analogous to antibody — establishing CRP as an "ante-antibody" opsonin that links the acute-phase response directly to complement effector mechanisms described in Complement Effector Functions and Regulation, and explaining why CRP-coated bacteria show enhanced phagocytosis even in the total absence of specific antibody.
Worked examples
Example 1 — Interpreting a clinical acute-phase panel. A 68-year-old man presents with fever, cough, and pleuritic chest pain. Labs return:
| Marker | Patient value | Typical reference range |
|---|---|---|
| CRP | 210 mg/L | <10 mg/L |
| Fibrinogen | 6.8 g/L | 2–4 g/L |
| Albumin | 24 g/L | 35–50 g/L |
| Serum iron | 25 μg/dL | 60–170 μg/dL |
| Ferritin | 480 ng/mL | 20–250 ng/mL |
| Hemoglobin | 11.2 g/dL | 13.5–17.5 g/dL |
Reasoning: CRP is elevated more than 20-fold — consistent with a major acute-phase response, typical of bacterial pneumonia (major CRP responses are more characteristic of bacterial than many viral infections, though this is a generalization with exceptions). Fibrinogen is elevated roughly 2-fold, the expected magnitude for a moderate-response protein, and contributes to the classically observed elevated erythrocyte sedimentation rate (fibrinogen promotes rouleaux formation). Albumin is low — a negative acute-phase protein being transcriptionally down-regulated in the liver, not (necessarily) evidence of malnutrition, so it should not automatically prompt nutritional intervention in this acute setting. Serum iron is low while ferritin (itself a positive acute-phase protein, and also an iron-storage protein) is elevated — this pattern is the signature of hepcidin-mediated iron sequestration: iron is being trapped intracellularly (raising ferritin, an iron-storage marker) rather than lost from the body, distinguishing it from true iron-deficiency anemia, where ferritin would be low. The mildly low hemoglobin is consistent with early anemia of inflammation. This panel, taken together, is exactly the expected constellation for an acute systemic response to bacterial infection driven by an IL-6 surge, and illustrates why ferritin must be interpreted alongside CRP — an elevated ferritin during active inflammation does not reliably indicate iron overload.
Example 2 — Distinguishing IL-6-dependent from IL-6-independent fever using a hypothetical dosing experiment. Suppose you inject three groups of mice (n = 10 per group) with:
- Group A: recombinant IL-1β alone
- Group B: recombinant IL-6 alone
- Group C: recombinant IL-1β plus an anti-IL-6 receptor antibody
You measure core temperature over 6 hours and plasma serum-amyloid-A (SAA, the mouse major acute-phase protein) at 24 hours. Expected results based on established biology: Group A (IL-1β alone) shows a robust fever response (core temperature rise of roughly 1.5–2°C, peaking at 2–4 hours) because IL-1β acts directly on hypothalamic endothelium to induce COX-2/PGE2, but shows only a modest SAA rise, because IL-1β is a comparatively weak direct hepatic acute-phase inducer on its own. Group B (IL-6 alone) shows a smaller, more delayed fever response (IL-6 is a weaker independent pyrogen than IL-1β/TNF-α, though not inert), but a large SAA rise, reflecting IL-6's dominant role in hepatocyte STAT3 signaling. Group C (IL-1β plus anti-IL-6R) shows a fever response similar in magnitude to Group A (because IL-1β's direct hypothalamic action does not require IL-6 signaling) but a substantially blunted SAA rise compared to IL-1β alone, because IL-1β's usual amplification of hepatic acute-phase genes proceeds largely through inducing endogenous IL-6, which is now receptor-blocked. This dissection — separable pyrogenic and hepatic-inducing activities, and the specific IL-6-dependence of the hepatic arm — is exactly the logic used in real experiments (and mirrored clinically by tocilizumab-treated patients, who mount fevers less reliably tracked by CRP).
Common misconceptions & traps
| Misconception | Correction |
|---|---|
| "Fever is a malfunction of thermoregulation caused by the pathogen." | Fever is an actively generated, centrally regulated host response (a reset thermostat), driven by the host's own cytokines and PGE2 — not a passive breakdown, and it can occur without any pathogen at all (e.g., in sterile tissue injury or malignancy). |
| "All acute-phase proteins increase during infection." | Negative acute-phase proteins (albumin, transferrin, transthyretin) decrease; a low albumin during acute illness reflects transcriptional reprioritization, not necessarily malnutrition. |
| "CRP tells you the specific pathogen or exact diagnosis." | CRP is a nonspecific marker of the magnitude of the IL-6-driven inflammatory response; it rises in trauma, autoimmune flares, and some cancers, and is generally (not absolutely) higher with bacterial than viral infection. |
| "A high fever itself is dangerous and must always be aggressively suppressed." | Fever within the typical range (up to ~40°C) is generally an adaptive, coordinated defense; routine antipyretic use in mild infections is a symptomatic-comfort choice, not always a physiological necessity, though very high or persistent fevers require clinical attention. |
| "Iron and ferritin move in the same direction, so low iron always means low ferritin." | In the acute-phase response, serum iron falls while ferritin rises — hepcidin traps iron intracellularly, and ferritin itself is a positive acute-phase reactant, so ferritin can be misleadingly "normal" or high even in true iron deficiency during inflammation. |
The scientist behind it
The modern molecular understanding of the acute-phase response owes much to Heinz Baumann and Jack Gauldie, whose collaborative work through the 1980s defined which cytokines induce which hepatic acute-phase genes and established IL-6 as the central hepatocyte-directed signal — work that reframed a century of purely descriptive clinical observation (fever, elevated sedimentation rates, "reactive" serum proteins) into a defined signaling pathway. Equally central is Charles Dinarello, who purified and characterized human IL-1 as an endogenous pyrogen in the early 1980s, providing the molecular identity behind Beeson's decades-old observation of a heat-labile leukocyte-derived fever factor, and who went on to define much of the biology of IL-1 family cytokines in inflammation broadly.
Origins · 1 minCharles A. Dinarello and how this was discovered
Where this matters today
Clinical diagnostics. CRP and, in some settings, procalcitonin are among the most widely ordered blood tests worldwide, used to triage suspected bacterial infection, monitor response to antibiotics, and track disease activity in autoimmune conditions such as rheumatoid arthritis and inflammatory bowel disease. Serial CRP trends often guide decisions about antibiotic duration and detect post-surgical complications before overt clinical signs appear.
Cardiovascular risk stratification. Chronically elevated high-sensitivity CRP (hs-CRP), reflecting low-grade vascular inflammation, is used as an adjunct marker of cardiovascular risk; the JUPITER trial demonstrated that statin therapy guided partly by elevated hs-CRP (in patients with normal LDL cholesterol) reduced cardiovascular events, supporting a causal — not merely correlative — role for chronic low-grade inflammation in atherosclerosis.
Anemia of chronic disease. Persistent IL-6-driven hepcidin elevation in chronic infections, autoimmune disease, and cancer causes a functional iron-deficiency anemia despite adequate or even excess body iron stores — understanding hepcidin biology has led to therapeutic hepcidin antagonists and anti-IL-6 strategies being trialed for this condition.
Cytokine-targeted therapeutics. Drugs blocking IL-6 (tocilizumab, sarilumab), IL-1 (anakinra, canakinumab), and TNF-α (infliximab, etanercept, adalimumab) — originally developed for autoimmune disease — directly dampen the acute-phase response and are used to treat cytokine release syndrome after CAR-T cell therapy and severe COVID-19, where an overwhelming, dysregulated acute-phase/cytokine response (a "cytokine storm") itself causes organ damage.
Amyloidosis. Chronic, unresolved acute-phase activation (e.g., in chronic infections, familial Mediterranean fever, or long-standing rheumatoid arthritis) can lead to persistent serum amyloid A elevation and its pathological deposition as AA amyloid in kidney, liver, and spleen — a direct clinical consequence of prolonged acute-phase signaling, treatable in part by suppressing the driving cytokine (commonly with IL-1 blockade in familial Mediterranean fever).
Veterinary and agricultural medicine. Acute-phase proteins (haptoglobin, serum amyloid A) are used as herd-health biomarkers in cattle and pigs to detect subclinical infection and evaluate transport stress, informing livestock management well before behavioral signs of illness appear.
Ask the tutor: How does the acute-phase response differ mechanistically between a bacterial infection and a sterile injury like major trauma or surgery?
Why does the body bother raising its temperature — isn't that metabolically costly?
Yes, fever is metabolically expensive — each 1°C rise increases basal metabolic rate by roughly 10–13%. But experimental and clinical evidence supports it as an adaptive trade-off. Comparative studies across ectotherms (lizards that behaviorally raise body temperature when infected) show that blocking fever (by preventing access to a heat source) increases mortality from bacterial infection, and similar impaired-fever-worse-outcome patterns appear in some mammalian and human studies with certain infections. Elevated temperature directly impairs replication of several temperature-sensitive pathogens, enhances neutrophil and lymphocyte motility and function, and improves antigen presentation. The response is centrally regulated (a reset "set point," not passive overheating), which is itself evidence it is a controlled, selected-for defense mechanism rather than a side effect.
Is CRP itself an antibody or antibody-like molecule?
No — CRP is not an immunoglobulin and is not generated by clonal selection or somatic recombination. It is a pentameric plasma protein, structurally a member of the pentraxin family, synthesized constitutively by hepatocytes and dramatically upregulated by IL-6 signaling. Functionally, though, it behaves as an "ante-antibody": once bound to phosphocholine on bacterial cell walls or on damaged host membranes, it can recruit C1q and activate the classical complement pathway, and it can also opsonize targets for Fcγ-receptor-bearing phagocytes. It provides pattern-based, innate opsonization available immediately, in contrast to antibody, which requires days for clonal expansion.
Why does albumin fall during the acute-phase response, and does that mean the patient is malnourished?
Albumin falls for at least two reasons: hepatocyte transcriptional and translational machinery is redirected toward positive acute-phase protein mRNAs (a relative, not absolute, shortage of capacity), and increased vascular permeability during inflammation allows albumin to leak into the interstitium, lowering the intravascular concentration further. Because this drop happens within a day or two — far faster than diet-related protein depletion could plausibly cause — a low albumin during acute illness is a poor marker of nutritional status and should not by itself trigger aggressive nutritional intervention; it normalizes as the inflammatory trigger resolves.
How does hepcidin actually block iron availability to bacteria?
Hepcidin is a small peptide hormone, induced by IL-6 acting on hepatocytes via STAT3, that binds the sole known cellular iron exporter, ferroportin, found on macrophages, duodenal enterocytes, and hepatocytes. Hepcidin binding triggers ferroportin internalization and lysosomal degradation, trapping iron inside these cells rather than releasing it into plasma. Since many extracellular bacteria depend on plasma-transferrin-bound iron for growth, this lowers the iron available to them — a strategy termed "nutritional immunity." The cost is that when the trigger (chronic infection, autoimmune disease) is prolonged, this same mechanism causes the anemia of chronic disease, since erythroid precursors are also denied adequate iron.
Do all species show the same acute-phase proteins in the same way?
No — this is an important simplification to flag. Humans and rabbits show dramatic (100- to 1000-fold) CRP responses, making CRP clinically useful in those species; mice show only a modest CRP rise and instead rely more heavily on serum amyloid P and serum amyloid A as their major acute-phase reactants, which is why mouse studies of "CRP biology" often actually measure SAA or SAP as functional analogs. Cattle and pigs rely heavily on haptoglobin and serum amyloid A as veterinary biomarkers. The overall cytokine logic (IL-6/STAT3 driving hepatic reprogramming) is conserved, but which specific protein serves as the dominant clinical marker varies by species.
How is the acute-phase response different from local inflammation, given they share cytokines?
Local inflammation is the tissue-level program at the site of infection or injury: vasodilation, increased vascular permeability, and leukocyte recruitment, driven by cytokines and chemokines acting on nearby cells within roughly millimeters. The acute-phase response is what happens when some of those same cytokines — especially IL-6 — escape into the systemic circulation in sufficient concentration to act as hormones on distant organs (liver, hypothalamus, bone marrow) that were never themselves infected. IL-6 knockout mice make this distinction empirically: they retain largely normal local inflammatory responses (leukocyte recruitment, local cytokine production) but show a severely blunted systemic hepatic acute-phase protein response, showing the two programs are mechanistically separable even though they share upstream triggers.
Why does IL-6 receptor blockade (tocilizumab) make it harder to detect infection in patients?
Because CRP synthesis is downstream of, and largely dependent on, IL-6 signaling through hepatocyte JAK/STAT3, blocking the IL-6 receptor removes the principal signal driving CRP production — even if the patient has an active, serious infection generating plenty of IL-6 upstream. Clinicians have observed that tocilizumab-treated patients can develop severe or even fatal infections with only mild or absent elevation in CRP and reduced fever, because the very biomarkers usually used to flag infection are pharmacologically silenced. This is a clinically important, well-documented pitfall: infection surveillance in these patients must rely more heavily on clinical exam, imaging, and other markers rather than CRP or temperature alone.