A unifying biologic explanation for “high-sensitivity” C-reactive protein and “low-grade” inflammation



Patients with chronic inflammatory conditions such as rheumatoid arthritis and systemic lupus erythematosus are at increased risk for cardiovascular disease. Although the mechanisms that underlie this association are unknown, interest has focused on the possible role of C-reactive protein (CRP) (1). Epidemiologic studies show that minor degrees of serum CRP level elevation are present in a high proportion of the population (2), are associated with varied lifestyles and medical conditions (3), and are associated with future risk of atherosclerotic events. In addition, a new, poorly defined term, “low-grade inflammation,” has been widely employed, and minor CRP level elevation is regarded as a marker of low-grade inflammation.

We suggest a unifying biologic explanation for these associations based on recent insights arising from 2 areas of scientific inquiry: the molecular mechanisms of innate immunity that trigger acute inflammation, and the unfolded protein response (UPR) to cell stress (4, 5). Although of interest to many medical specialties, such an explanation should be particularly welcome to rheumatologists, who are often called on to assist in interpreting reports of CRP concentrations.

Are high-sensitivity CRP and CRP different?

Following initial reports that minor CRP level elevation was associated with future cardiovascular events, highly sensitive assays were developed in order to accurately quantitate the low concentrations of CRP found in healthy people.

Subsequent publications referred to CRP concentrations determined by this method as high-sensitivity CRP (hsCRP) (6). Unfortunately, the misleading impression has arisen that hsCRP is different in some way from the CRP that we have long been familiar with. Thus, for example, we have heard a physician say, “A protein called high-sensitivity CRP predicts myocardial infarction.” Studies have even been published comparing levels of CRP and hsCRP (7). As Casas et al have recently pointed out, “It is important to recognize that … hs-CRP is just CRP itself, not anything new or different and in particular not a novel analyte with any special relationship to cardio-vascular disease” (8).

What is a truly normal CRP level?

The high prevalence of minor CRP level elevation has led to considerable uncertainty as to what to regard as normal CRP levels. Clinical laboratories avoid the term “normal values” and provide “reference values” instead, since the term “normal” has many possible meanings (9). As employed in clinical medicine, “normal” generally means an innocuous laboratory finding, i.e., one that the physician does not need to worry about.

Population studies over the last 4 decades have consistently shown a non-Gaussian, highly skewed distribution of CRP values (Figure 1), with no sharp cutoff between normal and elevated CRP levels. In recent studies, approximately 70–90% of samples from reference populations had CRP concentrations less than 3 mg/liter, suggesting that truly normal CRP levels in the American population are less than 2 or 3 mg/liter.

Figure 1.

The first report of a population study of C-reactive protein (CRP) values. A, Newborns, B, Mothers after delivery, C, Schoolchildren, D, Adult blood donors. Reproduced, with permission, from Kindmark CO. The concentration of C-reactive protein in sera from healthy individuals. Scand J Clin Lab Invest 1972;29:407–11.

Definition of normal CRP levels on the basis of population studies is difficult. Women have higher CRP levels than men, CRP values increase minimally with age, values differ between ethnic groups, obesity raises CRP levels, and various genetic polymorphisms are associated with higher CRP levels (2, 3, 10). Minor inflammatory stimuli such as cigarette smoke and air pollution are pervasive, as is the widespread use of estrogen-containing medications that can also influence CRP levels (3).

Consequently, reference values for CRP levels (that often vary from one laboratory to another) cannot be taken at face value. CRP levels between 3 and 10 mg/liter, which we deem minor CRP level elevation, are found in nearly 30% of Americans (2). A consensus has arisen that values should be greater than 10 mg/liter to reflect clinically significant inflammation (10–12).

Molecular mechanisms that trigger acute inflammation

Clinicians have long recognized the presence of inflammation by detecting functional or cellular elements of the inflammatory response: pain, warmth, erythema, swelling, and accumulation of leukocytes. Full-blown acute inflammation, in which the classic signs of inflammation are manifested and a substantial CRP response is seen, occurs in 3 basic situations: 1) infection, 2) tissue injury, and 3) untriggered spontaneous episodes that occur in the relatively uncommon autoinflammatory diseases (13). Infection triggers acute inflammation through 2 major mechanisms: binding to innate immune receptors by components of an invading pathogen, now referred to as pathogen-associated molecular patterns, and release of virulence factors that can cause tissue injury, with consequent liberation of endogenous products of damaged cells, i.e., danger (more appropriately damage)–associated molecular patterns (DAMPs) (4). The latter are sometimes referred to as alarmins (14). The mechanisms by which tissue injury elicits an inflammatory response are not as well understood, but are probably mediated largely through DAMPs.

The major purpose of these two types of inflammation appears to be to defend against microorganisms, remove necrotic tissue, and mediate adaptation and tissue repair. In the autoinflammatory diseases, in contrast, genetically based dysregulation of suppressive components of the inflammatory response results in purposeless episodes of inflammation.

What is meant by low-grade inflammation?

Modern understanding of the cellular and molecular mechanisms that underlie inflammation dictates that the classic definition of inflammation, i.e., the response to tissue injury, be revised. The presence of inflammation is now frequently inferred when increased concentrations of inflammatory mediators (such as inflammatory cytokines) or activation of transcription factors such as NF-κB are found (5).

Accordingly, a variety of new, poorly defined terms, such as “low-grade inflammation,” “mini-inflammation,” “subclinical inflammation,” “microinflammatory state,” and occasionally “chronic inflammation,” have been employed to indicate conditions where such changes are found in the absence of the classic signs of inflammation. Examples include obesity, insulin resistance, and obstructive sleep apnea. Such conditions tend to be associated with morbidity, implying the presence of some degree of metabolic malfunction. Minor CRP elevation is commonly regarded as indicating the presence of low-grade inflammation.

Low-grade inflammation differs from acute inflammation not only in magnitude, but in several other important ways as well: in underlying causes, apparent purpose, and molecular triggering mechanisms. These differences are striking enough so that two leading researchers in the field have suggested a distinct nomenclature for this state; both “para-inflammation” and “meta-flammation” (metabolically-triggered inflammation) have been proposed (4, 5).

Low-grade inflammation generally presents as a chronic condition in which the classic clinical signs of inflammation are lacking and only minor CRP level elevation is seen. It differs from acute inflammation in its underlying conditions, occurring in a variety of metabolic disorders characterized by cell stress and metabolic malfunction (4, 5), and is triggered by these processes. It differs in its apparent purpose; a major function of such inflammation appears to be the restoration of tissue homeostasis.

Molecular mechanisms that trigger low-grade inflammation

Macrophages and other cell types are capable of monitoring the homeostatic status of tissues and can detect changes from the optimal internal environment that result in cellular stress. Since such stress may lead to tissue malfunction, adaptive changes are required that involve inflammatory pathways (4). The mechanisms by which these changes are brought about are just beginning to be understood, and one such mechanism has been elucidated in detail: the UPR.

A thorough review of the molecular mechanisms of the UPR is beyond the scope of this article, but several excellent reviews have been published (15, 16). Briefly, the UPR occurs in response to endoplasmic reticulum (ER) stress, i.e., the presence of misfolded or unfolded proteins. The primary purpose of the UPR is to protect the ER from the toxic effects of such misfolded proteins and to target them for degradation. Many metabolic stressors can create ER stress, including glucose deprivation, perturbations of intraluminal calcium levels, cytokines, altered cellular redox state, hypoxia, toxins, viruses, increased protein trafficking, and nutrient excess or deficiency (5).

In response, the UPR is activated, mediated through 3 ER-resident transmembrane protein sensors: inositol-requiring enzyme 1 (IRE-1), PKR-like endoplasmic reticulum kinase (PERK), and activating transcription factor 6 (ATF-6) (Figure 2). These ER stress proteins are normally maintained in an inactive state through association with the ER chaperone BiP. In response to ER stress, however, BiP preferentially binds to and is sequestered by unfolded or misfolded proteins, with consequent release and activation of the 3 sensors. Activation of IRE-1, PERK, and ATF-6 in turn leads to reduced general protein translation and increased expression of a selected set of ER-associated proteins that assist in escorting misfolded proteins to degradation.

Figure 2.

The mammalian unfolded protein response (UPR) pathways. The endoplasmic reticulum (ER) chaperone BiP normally binds to and inhibits the 3 stress sensors: PKR-like endoplasmic reticulum kinase (PERK), inositol-requiring enzyme 1α (IRE1α), and activating transcription factor 6 (ATF6). In response to ER stress, BiP binds to unfolded or misfolded proteins, resulting in activation of the 3 stress sensors. Activated PERK then phosphorylates the translation initiation factor eukaryotic initiation factor 2α (eIF2α), decreasing general translation initiation. A few selected messenger RNAs (mRNAs) are preferentially translated, including ATF4, which in turn induces several UPR target genes. IRE1α activation leads to generation of the potent transcription factor X-box binding protein 1 (XBP1), which also induces several UPR target genes. Activation of ATF6 causes it to translocate to the Golgi, where it is cleaved to the active fragment ATF6/p50. The latter, after migration to the nucleus, induces transcription of a set of UPR target genes. S2P = site 2 protease; S1P = site 1 protease. Reproduced, with permission, from ref.15.

Recent studies have shown that the UPR can also lead to the release of inflammatory mediators in some cells, including macrophages, hepatocytes, adipocytes, and oligodendrocytes (5, 15), which appear to function as monitors, prepared to signal the presence of cellular stress. In such cells, the UPR leads to activation of inflammatory pathways signaled by IKK and JNK and mediated by the transcription factors NF-κB and activator protein 1. Both are critical to the induction of many inflammation-associated genes, including cytokines capable of acute-phase protein induction (Figure 3). In addition, accumulation of misfolded proteins in the ER causes calcium to leak from the ER, with generation of reactive oxygen species (15). Finally, activation of the transcription factor hepatocyte-specific CREB (structurally similar to ATF-6) ensues in the liver, with induction of CRP and of hepcidin, another acute-phase reactant, even in the absence of interleukin-6 stimulation (15, 17).

Figure 3.

Inflammatory pathways activated by the unfolded protein response. PERK and inositol-requiring enzyme 1α (IRE1α) are activated by endoplasmic reticulum stress, as shown in Figure 2. The attenuation of translation resulting from phosphorylation of eukaryotic initiation factor 2α (eIF2α) by activated PERK results in relative decrease of IκB, with consequent translocation of NF-κB to the nucleus. In addition, IRE1α activation leads to recruitment of tumor necrosis factor receptor α–associated factor 2 (TRAF2). The IRE1α–TRAF2 complex can then activate JNK and IKK, with consequent phosphorylation of the transcription factor activator protein 1 (AP1) and activation of NF-κB. Both NF-κB and AP1 induce transcription of the genes involved in the inflammatory response. Reproduced, with permission, from ref.15.

In summary, CRP induction does not necessarily require tissue injury and a classic inflammatory response. Rather, it may occur under conditions in which there are cells that are merely metabolically disturbed or stressed, and can be mediated by the resulting UPR.

Why does minor CRP level elevation predict coronary events?

While it is clear that minor CRP level elevation is associated with a moderately increased relative risk of future myocardial infarction (18, 19), many studies have shown that CRP determination adds little to a thorough evaluation of known risk factors (12, 20, 21). Risk factors are merely epidemiologically determined associations. In many cases, such as low levels of physical activity, the mechanistic explanation for their association with atherosclerosis is unknown. It is generally assumed that they are causal, inducing altered physiology or disturbed metabolism in some way.

In an informative study, 9 risk factors were found to account for almost all of the population-attributable risk of myocardial infarction: dyslipidemia, smoking, hypertension, diabetes mellitus, abdominal obesity, depression and other psychological factors, and low levels of physical activity, of fruit and vegetable consumption, and of alcohol consumption (22). Most, if not all, of these risk factors may be presumed to represent some degree of metabolic perturbation. Indeed, all have been reported to be associated with minor CRP level elevation (3). It is reasonable to conclude that CRP predicts coronary events merely because it is associated with, and presumably reflects, these risk factors. Rather than being thought of as an effector, CRP must therefore be regarded as a reflector, informing us of the presence of stressed cells.

The possibility that CRP contributes directly to the pathogenesis of atherosclerosis has been controversial. Arguments have been presented both supporting and refuting this view (8). Although the evidence cited above suggests that CRP is a reflector, it is still possible that CRP may contribute to the pathogenesis of atherosclerosis, although recent Mendelian randomization studies argue against this possibility (23).


We do not offer an opinion about the clinical utility of CRP determination to estimate the future risk of atherosclerotic events. Rather, we attempt to understand the mechanisms by which CRP is induced in the many conditions regarded as examples of low-grade inflammation. The UPR, which occurs in response to cell stress and metabolic malfunction, can in some cell types lead to CRP induction. Therefore, modest CRP level elevation can signal the presence of cell stress rather than indicating inflammation as we have traditionally understood it. These findings provide a molecular explanation for the minor CRP level elevation seen in apparently noninflammatory conditions that are regarded as examples of low-grade inflammation, including the risk factors for atherosclerosis, and provide a perspective with which to interpret the minor degree of CRP level elevation that is seen in a broad variety of clinical states.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Kushner had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Kushner, Samols, Magrey.

Acquisition of data. Kushner, Samols, Magrey.

Analysis and interpretation of data. Kushner, Samols, Magrey.


We are grateful to Dr. John Goldman of Atlanta for suggesting that this article be written, and to Drs. Leonard Calabrese, Brian Mandel, and Sylvie Hauguel-De Mouzon for helpful suggestions.