Circadian rhythms in rheumatoid arthritis: Implications for pathophysiology and therapeutic management

Introduction

It is a well-known phenomenon that disease-related symptoms in patients with chronic inflammatory diseases exhibit circadian rhythms. This has been studied in, e.g., patients with rheumatoid arthritis (RA) (1–7). It is the clinical experience of rheumatologists that RA patients particularly experience joint pain, morning stiffness, and functional disability in the early morning hours. It is also remarkable that these diurnal variations demonstrate large amplitudes, with the patient's condition being poor in the early morning and disease activity being mild or moderate in the early evening. This has been demonstrated using grip strength as a parameter; grip strength was, on average, 233 mm Hg at 6:00 AM and 297 mm Hg at 6:00 PM, an increase of 27.5% (3). It has been further exemplified by overall pain reported using a 0–10-point visual analog scale, with an average pain score of 6.3 at 8:00 AM and 4.5 at 6:00 PM (decrease of 28.6%) (8). Results of assessment of symptoms in large clinical trials can vary markedly depending on the time of day that the visit takes place.

The causal mechanisms underlying this pronounced diurnal cycle of changes in symptoms with a maximum in the morning are relevant to the pathophysiology of RA (9, 10), to clinical patient care, and to optimization of treatment strategies (11). Since circadian rhythm is generated solely in the higher brain centers of the hypothalamus, learning from these experiments of nature can provide new clues to understanding neuroendocrine–immunologic pathways relevant to rheumatic diseases.

This review briefly summarizes present knowledge on how circadian rhythm is generated in the central nervous system (CNS) and addresses studies demonstrating connections between the central oscillator and neuroendocrine pathways and the periphery (peripheral inflammation). Examples of circadian rhythms of hormones, specifically in relation to RA, are discussed, as is the circadian rhythm of pain, stiffness, and functional disability in health and disease. Circadian rhythms of RA-relevant immunologic parameters, including linkage between the diurnal variation of pain (and other RA symptoms and comorbidities) and the cycle of hormonal, neuronal, and immunologic parameters, are considered. Finally, important implications regarding therapy are derived, which in the future might lead to optimized treatment of RA and other chronic inflammatory diseases. A methodologic appendix (Appendix A) explains how several studies were mathematically analyzed and their data combined.

Generation of the circadian rhythm and modulation of peripheral target tissues

In the early 1970s, brain lesion experiments and metabolic and electrophysiologic studies indicated that in mammals the central circadian oscillator is located in the hypothalamic suprachiasmatic nucleus (SCN) (12, 13). Lesions in this area of the brain led to complete loss of circadian rhythm, and SCN transplants restored diurnal rhythmicity (14, 15). Further studies demonstrated that neurons of the SCN contain a genetically driven clock mechanism (16, 17). In the early 1980s, CLOCK genes were identified in Drosophila, and 10 years later, homolog genes were identified in mammals (18, 19). Examples of CLOCK genes are, among others, period homolog 3 (Per3; chromosome 1p36.23), period homolog 2 (Per2; chromosome 2q37.3), NR1D2 (Rev-Erbα; chromosome 3p24.2), BMAL (Arntl; chromosome 11p15), cryptochrome 2 (Cry2; chromosome 11p11.2), timeless (chromosome 12q12–q13), cryptochrome 1 (Cry1; chromosome 12q23–q24.1), and period homolog 1 (Per1; chromosome 17p13.1–17p12). These CLOCK genes and their products exhibit a transcriptional–translational feedback loop that guarantees a nearly 24-hour cycle (17, 20).

Figure 1A demonstrates the orchestrated activities of the main players in the SCN, and Figure 1B shows the target areas in the CNS. The SCN connects to other brain centers in the CNS. From these brain centers, there are connections to the non-CNS periphery. The connections are made possible by secreted hormones and neuronal pathways, e.g., the sympathetic nervous system (SNS) (Figure 1B).

Figure 1. Molecular control of intracellular CLOCK pathways and target systems in the brain and periphery. A, The 2 important parts of the suprachiasmatic nucleus (SCN), the core neurons (left) and the shell neurons (right). The core neurons receive direct neuronal inputs from the retina via the retinohypothalamic tract. Glutamate, substance P (SP), and pituitary adenylate cyclase–activating polypeptide (PACAP) are the main neuropeptides in this interaction. These neuropeptides signal via, e.g., cAMP, protein kinase A (PKA), and CREB to induce the transcription of Per1 mRNA and Per2 mRNA via the cAMP response element (CRE). Per1 and Per2 induce SCN core neuron activation and local regulation of shell neurons via gastrin-releasing peptide (GRP), vasoactive intestinal peptide (VIP), γ-aminobutyric acid (GABA), and SP. Via the same neuropeptides, coordination of other central nervous system (CNS) targets takes place. Shell neurons have their own endogenous 24–25-hour rhythm, which is generated by a transcriptional–translational feedback loop (72). This feedback loop is generated by CLOCK (C) and BMAL (B), which dimerize and bind to E-boxes in the promoter region of period (Per) genes, cryptochrome (Cry) genes, Rev-Erbα, and BMAL, activating their transcription. The corresponding proteins accumulate in the cytoplasm, translocate to the nucleus, and inhibit binding of CLOCK and BMAL to the E-boxes (the feedback). This inhibition leads to down-regulation of transcription of these genes. A very similar loop exists with respect to Rev-Erbα and BMAL regulation. The degradation process via ubiquitination and SUMOylation leads to the known 24–25-hour cycle. In a certain activation state of SCN shell neurons, they can modulate remote neurons in the hypothalamus and elsewhere via vasopressin and GABA. The self-sustained rhythmicity of the shell neurons is synchronized with the external light/dark cycle via the influence of the above-mentioned input from core neurons. B, Targets of the SCN core and shell. The targets include, among others, sleeping, arousal, and feeding centers, other nuclei in the hypothalamus and the pineal gland, and higher centers of the sympathetic nervous system (SNS). The interaction of the SCN neurons with these areas leads to the known circadian rhythmicity of many different neuroendocrine systems, which transmit their signals by hormones and neurotransmitters to the periphery. CRH = corticotropin-releasing hormone; ACTH = adrenocorticotropic hormone; TRH = thyrotropin-releasing hormone; TSH = thyroid-stimulating hormone; GH = growth hormone; GHRH = GH-releasing hormone; GnRH = gonadotropin-releasing hormone; LH = luteinizing hormone; FSH = follicle-stimulating hormone; Ncl. = nucleus; ADH = antidiuretic hormone.

The earth's daily light/dark cycle is not required for circadian rhythms, because the shell neurons of the SCN (Figure 1A) demonstrate a self-sustained rhythmic loop with a period of ∼24 hours. The intrinsic rhythm in humans is ∼24.2 hours, and that in Drosophila is 23.7 hours. Circadian rhythm also occurs in fungi, bacteria, algae, worms, plants, and other organisms, with a similar period. Interestingly, even single cells in the human body, such as cardiomyocytes, peripheral blood mononuclear cells, natural killer cells, liver cells, and most other cells, generate a circadian rhythm of ∼24 hours, which is controlled by superordinate SCN activities via hormones and neurotransmitters released into the blood and locally into the vicinity of nerve terminals (21–24). Thus, most cells of the body, including immune cells, are controlled by the SCN, leading to coupled cell and organ activities in the periphery with a 24-hour daily cycle (20). Such coupling phenomena may be important for the antiinflammatory cooperation of hormones and neurotransmitters in RA, which has been demonstrated for cortisol and norepinephrine (25). Since RA-associated symptoms undergo a similar 24-hour cycle, leading to large variations in disease activity, this superordinate neuroendocrine center in the hypothalamus is of great importance with regard to RA pathophysiology.

Circadian rhythms of hormones

As early as the 1950s and 1960s, shortly after the discovery of cortisol, the circadian rhythm of this hormone was described (e.g., ref.26). The cycle of the hypothalamic–pituitary–adrenal axis (HPA axis) shows a maximum in the early morning hours at 8:00 AM and a nadir at midnight (Figure 2A). Interestingly, in healthy subjects, bone-resorbing activity is highest between 5:00AM and 7:00 AM, which conforms to cortisol and tumor necrosis factor (TNF)/interleukin-6 (IL-6) rhythms (27). Bone-resorbing activity has not been investigated in patients with RA.

Figure 2. Circadian rhythm of hormones in healthy subjects and in patients with rheumatoid arthritis (RA). A, Serum levels of cortisol. B, Serum levels of cortisol in RA patients with a high degree of inflammation and those with a low degree of inflammation, as determined by the erythrocyte sedimentation rate (ESR). C, Serum levels of melatonin. D, Serum levels of prolactin. The graphs in A, C, and D were constructed from data in refs.40,42,45,59, and73–77 (A),39,40, and78–82 (C), and41,42, and83 (D), using the methods described in Appendix A. Solid lines represent healthy subjects; dotted lines represent RA patients. The graph in B was constructed using the data of Neeck et al (28) after normalization to the 24-hour mean in each patient (n = 6 patients with high ESR and 6 patients with moderate ESR). Values are the mean and SEM. Horizontal bars starting at 21:00 (9:00 PM) indicate sleeping time.

The cortisol rhythm in patients with RA whose disease activity is relatively low to moderate does not differ from that in healthy subjects (Figures 2A and B). This is the case with regard to the period, the amplitude, and the time point of the minimum and peak of the cycle (Figure 2A). It has been nicely delineated that this rhythm can be highly disturbed in RA patients when disease is very active, leading to a flattening of the response curve (Figure 2B), and 2 peaks appear, in the morning and the afternoon (28, 29). The loss of circadian variation was confirmed in the animal model of adjuvant-induced arthritis (30–32). In addition, RA patients with high disease activity have elevated serum cortisol levels, which are, however, inadequately low in relation to ongoing inflammation (28, 29). Since cortisol is the strongest endogenous antiinflammatory substance, its up-regulation in the early morning is most probably related to inhibition of inflammation during the day, and its down-regulation during the evening and night is linked to an increase of inflammation during the early morning.

Besides cortisol, 2 other hormones, melatonin and prolactin, demonstrate a perfect 24-hour rhythm (Figures 2C and D). Both hormones have been linked to stimulation of the immune system, which would lead to an increase in proinflammatory conditions in RA (33–38). The typical circadian rhythm of melatonin exhibits a maximum at 3:00 AM, which is quite similar to that of prolactin (Figures 2C and D). When data from all available studies are combined, the rhythms of these 2 hormones are not shown to be markedly different in patients with RA as compared with healthy controls (see references in Figures 2C and D). However, one study demonstrated that serum levels of melatonin reached a peak ∼2 hours earlier in RA patients than in controls (39). In RA patients, melatonin levels exhibited a wide plateau lasting 2–3 hours, an effect not observed in healthy controls (39). After the peak was reached, melatonin levels decreased similarly in RA patients and healthy subjects (39). Furthermore, in a study of subjects from a northern European country, serum levels of melatonin appeared to be elevated in patients with RA as compared with controls (40). Similarly, Chikanza et al found that serum levels of prolactin during the night were significantly higher in RA patients compared with controls (41), and this has been confirmed by others (42).

Both elevated melatonin and elevated prolactin will probably establish a more proinflammatory environment exactly at the time point when cortisol levels are lowest. This is particularly true because the ratio of prolactin to cortisol peaks at 2:00 AM in RA patients (42). Both hormones, prolactin and melatonin, induce a Th1 immune response and may thus lead to an unwanted increase in related cellular immune phenomena in RA patients (for review, see. ref.43).

Since RA prevalence and incidence are higher in women than in men, the question of whether neuroendocrine mechanisms are different in women and men, and whether this may contribute to sex differences, might be important (44). In healthy subjects, the circadian rhythm of cortisol levels is similar in women as compared with men (45); a sex difference for prolactin and melatonin has not been investigated. Furthermore, there are no published reports of studies comparing the circadian rhythm of cortisol, melatonin, and prolactin levels separately in women and men with RA. This might be an interesting avenue for future research.

Circadian rhythms of pain thresholds and symptoms

Similar to hormone cycles, serum levels of β-endorphin, an endogenous μ-opioidergic analgesic derived from the common precursor hormone proopiomelanocortin (adrenocorticotropic hormone [ACTH] is also derived from this prehormone), demonstrate a minimum at approximately midnight (Figure 3A). Thus, the nadirs of β-endorphin, cortisol, and ACTH appear exactly at the same time, which can be important since glucocorticoids are also known to inhibit pain pathways (46). In healthy subjects, the pain threshold has a circadian rhythm, demonstrating a minimum at 1:00–3:00 AM (Figure 3B). This nadir is closely related to the nadir of endogenous analgesics and cortisol. Thus, in healthy subjects, responsiveness to painful stimuli is highest during the night. It is interesting that pain thresholds were generally lower in women as compared with men, as in a headache model in healthy subjects (47). Whether this is a factor in the female-to-male preponderance of RA or in the greater disease severity in women with RA is presently not known.

Figure 3. Circadian rhythm of A, serum β-endorphin levels and B, pain measures in healthy subjects, and of C, stiffness, D, pain measures, and E, functional disability in patients with rheumatoid arthritis. The graphs were constructed from data in refs.84 and85 (A),47 and85 (B),1,3,4,8, and86 (C),4–6,8, and86 (D), and4 and86 (E), using the methods described in Appendix A. Values are the mean and SEM. Vertical dotted lines show the peak or minimum value for the respective parameter. Horizontal bars starting at 21:00 (9:00 PM) indicate sleeping time.

Patients with RA exhibit maximum stiffness at 6:00 AM (Figure 3C), maximum pain at 8:00 AM (Figure 3D), and maximum functional disability at 6:00 AM (Figure 3E). Grip strength has been found to reach a minimum at 6:00–8:00 AM (2, 48). It is interesting that there is a time shift between healthy subjects and patients with RA with respect to pain: whereas pain levels demonstrate a maximum at 8:00AM in RA patients, in controls, pain thresholds, an inverse marker of pain levels, are lowest at 3:00 AM (lag phase of 5 hours), as indicated above (Figures 3B and D). Below, we will discuss whether the lag phase in RA may be related to increased cytokine curves. At present, there are no published data from studies comparing the circadian rhythm of symptoms such as stiffness, pain, functional ability, or grip strength separately in women and men with RA, which might be a factor in the greater severity of these problems in female patients.

Similar to RA, in other diseases such as myocardial infarction, angina pectoris, and renal colic, the peak frequency of painful attacks lies between 4:00 AM and 8:00AM (49–51). This indicates that similar phenomena might play a role in RA and in these diseases.

Circadian rhythms of immunologic parameters

Cytokines exhibit a marked rhythmicity, as demonstrated in Figure 4 for serum TNF and serum IL-6. In healthy subjects, the peak value of TNF is reached at ∼3:00 AM and that of IL-6 at ∼6:00 AM (Figures 4A and B). This nicely parallels the sequence of TNF-induced IL-6 secretion that is well known from in vitro studies. In addition, a similar sequence of IL-6 and IL-6–dependent secretion of fibrinogen has been documented, with maximum fibrinogen levels at 10:00 AM (52).

Figure 4. Circadian rhythm of serum tumor necrosis factor (TNF) and interleukin-6 (IL-6) levels in healthy subjects (solid lines) and in patients with rheumatoid arthritis (dotted lines). The graphs were constructed from data in refs.42,45,53–55,78, and87 (A) and11,40,45,52,53,59,76,79, and88–91 (B), using the methods described in Appendix A. Values are the mean and SEM. Horizontal bars starting at 21:00 (9:00 PM) indicate sleeping time.

In patients with RA, the peak level of TNF has been reported to appear at 6:00 AM and that of IL-6 at 7:00 AM (Figures 4A and B). Thus, for both cytokines, a time shift of the peak value toward the morning appears. In healthy subjects, serum TNF and IL-6 levels are ∼2–5 pg/ml whereas in RA patients these levels are 20–50 pg/ml. From this information, it becomes evident that the amplitude of the curve is much higher and the curve necessarily must be broadened for RA patients as compared with controls (Figure 4). In healthy subjects, serum levels of TNF and IL-6 have already begun to decrease at 6:00 AM and 9:00AM, respectively, whereas in RA patients these levels remain elevated until 10:00AM and 11:00AM, respectively. The fact that the amplitude is higher and the curve broadened for these proinflammatory cytokines in RA patients versus controls despite the similarity of the circadian curves for serum cortisol, with similar amplitude and shape, indicates inadequate cortisol secretion in relation to inflammation in RA, as mirrored by TNF and IL-6 release.

It is interesting that serum levels of IL-2 and interferon-γ (IFNγ) demonstrate peak levels at midnight–2:00AM in healthy subjects (53–55). These 2 cytokines, similar to TNF, induce a Th1 immune response, as has also been reported for growth hormone and prolactin (see above). Thus, particularly during the night, Th1 immune responses develop with a preponderance of cellular immunity (10, 56). Since cortisol and norepinephrine (the SNS) would, in contrast, support a Th2 immune response (57), the levels of these factors are particularly low between 11:00 PM and 5:00AM. At present, the circadian rhythm of IFNγ and IL-2 in patients with RA is not known.

In addition, a circadian rhythm exists for immunoglobulins in RA, which has nicely been demonstrated for IgA rheumatoid factor (peak at 8:00AM) and IgM rheumatoid factor (peak at 2:00AM) (5). In addition, circulating immune complexes exhibit a circadian rhythmicity in RA, with a peak between 6:00AM and 9:00AM (58). This demonstrates that 2 other important disease-relevant parameters display an elevation in the early morning hours in patients with RA.

Links between hormones, immunologic activation, and pain pathways

Crofford et al demonstrated a lag time between increase in IL-6 levels and increase in cortisol levels (plus 60–120 minutes) or ACTH levels (plus 60 minutes) in patients with RA (59). The present systematic review of many independent studies confirms the occurrence of this lag phase between IL-6 increases and cortisol increases in both healthy subjects and RA patients (compare Figures 2A and 4B). Similarly, there is a lag time between increase in serum cortisol levels and increase in serum TNF levels (5 hours in healthy subjects, 2 hours in patients with RA) (compare Figures 2A and 4A). Thus, we assume that increases in cytokine levels during the early night drive the increase of cortisol secretion and, most probably, also the activity of the SNS (57, 60). In turn, these 2 systems may then inhibit increased cytokines in the morning hours and during the day. On the other hand, around midnight, the decreases in levels of cortisol and β-endorphin, both of which inhibit secretion of IL-6, TNF, and other cytokines (61), together with the increases in levels of melatonin, growth hormone, and prolactin, drive nocturnal increases of TNF, IFNγ, IL-2, IL-12, and IL-6 (10, 34, 40). In summary, we have found a well-orchestrated up- and down-regulation of cytokines and hormones which, most probably, lead to up- and down-regulation of peripheral immune responses.

It is reasonable to suggest that increasing levels of RA-relevant cytokines such as TNF, IFNγ, IL-2, and IL-6 drive the local proinflammatory process in joints and secondary lymphoid organs in the early morning hours. Increased inflammatory conditions stimulate edema formation via bradykinin/prostaglandins/substance P and pain sensitization (62, 63). In addition, recent research has demonstrated that proinflammatory cytokines drive pain pathways at the spinal level (64). At present, we do not know whether spinal cytokine levels demonstrate a similar rhythmicity, but if peripheral cytokines can translocate or transmit the signal to the CNS, one would expect the occurrence of pain sensitization and pain augmentation in the early morning hours. These unwanted side effects of elevated circulating proinflammatory cytokines ultimately lead to symptoms of stiffness (edema formation), pain, and functional disability (edema formation) in the morning. In addition, TNF and IL-6 play an important role in the worsening of symptoms of several RA-related comorbidities, such as cardiovascular disease, osteoporosis, depression, and sleep disturbances, all of which demonstrate maximum severity during the night and in the early morning hours (27, 49, 65).

It is tempting to speculate that coupling of important neuroendocrine immunologic mediators (HPA axis together with SNS) and uncoupling of other factors (HPA axis/SNS versus prolactin, melatonin, growth hormone, TNF, IL-6, etc.) has been evolutionarily conserved to overcome infectious diseases, particularly, during the night, with the intention that energy resources can be allocated to an activated immune system (66). The immune system needs large amounts of energy when in an activated state (67). It has been hypothesized that in RA this evolutionarily conserved mechanism is used in a nonspecific and unfavorable way (66).

In summary, an important finding in RA is increased secretion of prolactin/melatonin and inadequately low secretion of cortisol; circadian curves of cortisol are very similar to those in healthy subjects (see above). Although one might expect an increased cortisol curve, in order to dampen exaggerated cytokine secretion, in RA, it seems that this is only partly possible in patients with very high disease activity (28, 29) (Figure 2B). Nevertheless, the somewhat higher cortisol levels in patients with very active disease are not sufficient to alleviate the disease process in RA. This inadequacy of cortisol secretion demonstrates that treatment with exogenous glucocorticoids at the beginning of RA, during disease flares, or during smoldering inflammation in mild-to-moderate RA can be viewed as a substitution therapy for the functionally disturbed HPA axis.

Implications for therapy

As early as 1989, it was suggested that an evening dose of glucocorticoids may alleviate early morning symptoms in patients with RA (9). However, evening treatment with glucocorticoids at doses of 10 mg prednisolone may not be sufficient to dampen the response in the early morning, due to the short half-life of prednisolone (2 hours) and the intense immune activation during the night. In a seminal study, Arvidson et al demonstrated that the administration of prednisolone at 2:00AM as compared with 7:30AM markedly reduced morning stiffness, pain, disease activity assessed by the Lansbury index (68), and serum IL-6 levels after only 5 days of treatment (11) (Figure 5). Those authors studied 2 groups of RA patients, who received 5.0–7.5 mg prednisolone at 2:00 AM (therapeutic group I; n = 13) or at 7:30 AM (therapeutic group II; n = 13). Baseline values in the 2 groups were very similar. Although the study did not include a placebo group or a crossover design, the results were quite remarkable (11) (Figure 5). No other studies to replicate these findings have been carried out, but the interest of pharmaceutical companies has been stimulated. Development of vehicles for medication delivery that allow timed release of glucocorticoids at 2:00AM might be a new treatment option in RA and other chronic inflammatory diseases.

Figure 5. Reduction of clinical symptoms and serum interleukin-6 (IL-6) levels after nighttime (2:00 AM) and morning (7:30 AM) administration of 5.0–7.5 mg prednisolone for 5 consecutive days in 2 groups of rheumatoid arthritis patients. The therapeutically induced decrease in each variable was calculated for every patient using the original data reported by Arvidson et al (11). In that study, baseline parameters and the number of patients receiving prednisolone were similar in both groups. Median values in the 2 groups of patients were compared by Mann-Whitney test, using SPSS, version 12.0 (SPSS, Chicago, IL). Values shown are the mean and SEM degree of change in each variable. VAS = visual analog scale.

The question arises as to why immunosuppressive treatment with glucocorticoids can better inhibit proinflammatory sequelae when given at 2:00AM (as TNF release is increasing) as compared with administration at 7:30AM (as TNF release is declining) (Figure 6). This is a very important consideration in the understanding of antiinflammatory counterregulation of immune responses. In a recent study in rats, it was demonstrated that inhibition of splenic TNF secretion occurred only if the immunosuppressive agent (released neurotransmitters of the sympathetic nerve fiber via the β-adrenoceptor) influenced TNF secretion at an early time point in relation to lipopolysaccharide (LPS) administration (i.e., as TNF release was increasing) (69). Later administration of the immunosuppressive agent, during the time TNF release was declining, did not affect TNF secretion (69). In that study, the time course of TNF release was very similar to the nighttime increase and subsequent decrease of TNF levels in healthy subjects and RA patients, which occurs over a period of 3–5 hours. The time curve demonstrated in Figure 4A is quite similar to the time curve in the spleen study in rats (69). This typical time curve of TNF increase and decrease has been documented in several ways (TNF messenger RNA rises after 30 minutes of exposure to a stimulus such as LPS, TNF protein rises after 1 hour, maximum TNF protein secretion occurs at 2 hours).

Figure 6. The effect of administration of immunosuppressive agents during periods of increasing and decreasing tumor necrosis factor (TNF) secretion. Data are derived from a perfusion study of the rat spleen (69), which exhibits a time course of TNF increase and decrease similar to that in humans. It appears that the turning-on phase of a proinflammatory response is more amenable to treatment with immunosuppressive agents such as glucocorticoids than is the turning-off phase; therefore, it is hypothesized that immunosuppressive therapy administered during the turning-on phase has a stronger capacity to inhibit proinflammatory sequelae.

It is has been demonstrated that glucocorticoids induce transcription of the IκBα gene, which results in an increased rate of IκBα protein synthesis and inhibition of proinflammatory NF-κB effects (70). Others have shown that glucocorticoids can interfere with the transcriptional activation potential of DNA-bound NF-κB complexes, leading to antiinflammatory effects (71). These effects appear very early in the turning-on phase of a proinflammatory response (Figure 6). We hypothesize that the turning-on phase of a proinflammatory reaction is much more amenable to immunosuppressive treatment than is the turning-off phase (Figure 6). It is suggested that regulation of an important proinflammatory factor such as TNF must happen very early, because otherwise an overwhelming secretion of this harmful cytokine may take place. In addition to understanding that cortisol may be secreted at low levels and therapeutically substituting this hormone in levels adequate to compensate for this, timing of glucocorticoid administration such that it is given during the turning-on phase of TNF secretion can be of major importance.

Based on the above-mentioned mechanisms, other therapeutic modalities with timed release of hormonal antagonists of melatonin or prolactin might be of interest. In addition, timed release of antiinflammatory β-adrenergic agonists or androgens (which show a circadian rhythm similar to that of cortisol) together with glucocorticoids might be another option. The spectrum of possible old drugs in a new form of drug delivery will probably lead to optimized drug release patterns with improved immunosuppressive activities. Stronger inhibition of nighttime proinflammatory cytokines such as TNF and IL-6 should lead to a reduction of RA-related symptoms in the early morning and possibly, over the long term, to fewer RA-related comorbidities such as cardiovascular disease, osteoporosis, depression, and sleep disturbances.

Conclusions

Since circadian rhythmicity of neuroendocrine pathways is tightly coupled to immune activation, we believe timed release of neuroendocrine factors represents an entire new field of RA management. Future studies in this direction will demonstrate whether timed release also leads to the ability to reduce drug dosages below critical levels at which adverse events appear. Since circadian rhythmicity is completely dependent on subordinate central nervous mechanisms (the SCN) and downstream hormonal and neuronal pathways, study of circadian coupling and uncoupling phenomena demonstrates once again that neuroendocrine factors play a critical role in the pathophysiology of chronic inflammatory diseases such as RA.

APPENDIX A

METHODOLOGY USED IN THE PRESENT REVIEW

PubMed (National Center for Biotechnology Information, National Institutes of Health) was screened for available literature on circadian rhythms of hormones, immunologic parameters, and variables of pain and functional disability. This search included literature on healthy subjects and patients with RA. Studies that compared controls with RA patients were the particular focus of the literature retrieval. Since units varied significantly among the different studies, for every reference the parameter of interest was recalculated as the percent of the mean of all values obtained during the entire day (unit: % of 24-hour mean). Such an analysis is typical for evaluation of circadian rhythm data. For parameters presented in the figures, the mean value was calculated including values from single studies obtained at specific time points during the day. Necessarily, the presented values undulate around 100% irrespective of whether the original data were presented in pg/ml, pmoles/liter, nmoles/liter, or another unit. We paid close attention to ensure that the respective time point during the day was identified in the various references and correctly assigned. In most cases, values for the same time point during the day from several studies were assessed; in these cases, the error bars depict the SEM from all available studies. However, if only 1 study was available for a certain time point, no error bars are shown for that time point.