MUMultiple speciesBmoderateT2limited target data
Serum amyloid A rose about 227-fold in an equine experiment, haptoglobin only 1.1-fold. How inflammation starts, how it ends, what CRP and SAA measure.
Executive summary
Inflammation is a regulated defence and repair response, not a defect. It has a beginning, a peak and an active ending. When that ending fails to come, chronic inflammation sets in. It is measured through acute phase proteins: in the horse mainly serum amyloid A, in human routine care C-reactive protein. Both show the scale and the course, but name neither the cause nor the site of the inflammation.
Key points
Inflammation is the coordinated response of a tissue to damage or to intruders. The four signs described since antiquity (redness, heat, swelling, pain) are the visible consequences of three processes: the vessels in the surrounding area become more permeable, immune cells move in, and signalling molecules hold that state in place until the trigger has been removed. The biological purpose is not destruction but containment: the damage is walled off, cleared away and then repaired.
Because without it there is no wound healing and no defence against infection. It turns harmful only when it is too strong, lasts too long, or runs in the wrong place. It is precisely these three ways of going wrong, and not inflammation itself, that lie behind most inflammatory disease patterns in horse and human.
From this follows a distinction that carries the whole of the rest of this text: high inflammatory activity is not a finding in itself. Only the course over time says whether a sensible response is running or whether it has derailed.
The innate immune system recognises two classes of signal: foreign patterns from bacteria, viruses and fungi, and the body's own alarm substances that leak out of damaged cells. Both bind to receptors on phagocytes. Some of these receptors switch on a protein complex, the inflammasome, which brings the messenger interleukin-1 beta into its active form. Together with tumour necrosis factor alpha and interleukin-6 it forms the trio that steers everything that follows.
Interleukin-6 leaves the site of inflammation and reaches the liver. There the liver cell switches its production programme: it makes less albumin and more of what are called acute phase proteins. Serum amyloid A is one of them, a small protein of 104 amino acids that has remained remarkably unchanged across the vertebrates and whose blood level can rise up to a thousandfold within a day. This evolutionary conservation is the real reason why a comparison between horse and human makes any sense at all.
Acute inflammation lasts hours to days and limits itself. In the equine experiment with artificially induced joint inflammation, the systemic signs had gone in most animals by the second day, and every blood marker was back at baseline on the fifteenth day. Chronic inflammation arises when the trigger persists or when the closing process fails to take hold.
In humans this permanent state in later life goes by the name inflammaging: a low-grade, lasting activation with no single identifiable trigger. A review of 23 intervention studies in older adults found that exercise programmes lower CRP, interleukin-6 and tumour necrosis factor alpha, the effect being clearer in healthy people than in those with an existing disease. A later analysis of randomised trials confirmed the reduction in CRP and interleukin-6, strongest when training was combined with additional protein intake.
The reason chronic inflammation is so hard to spot lies in the order of magnitude. An acute response lifts the reading by two to three powers of ten; a chronic one, by contrast, moves it only slightly beyond the normal range and holds it there. A single reading can barely tell that state apart from ordinary scatter. Only repeated measurements over weeks make it visible, and it is exactly such series of readings that are largely missing in the horse.
The subsiding of an inflammation was long taken to be a passive ebbing away, on the grounds that the signalling molecules break down. That picture is out of date. From polyunsaturated fatty acids the body builds a class of signalling substances of its own: lipoxins from arachidonic acid, and resolvins, protectins and maresins from the omega-3 fatty acids EPA and DHA. They stop the influx of further immune cells without weakening the defence against pathogens, and they speed up the clearance of dead cells by phagocytes.
How far this process can be timed is shown by a mouse experiment: a repeated, very low-dose mixture of five such substances shortened the window until the recruited immune cells had been cleared by more than 70 per cent. That is a finding in rodents in a peritoneal model, not a clinical statement, and it is confirmed neither for the horse nor for the human.
No. There is at present no field-ready measurement for resolution mediators in the horse. What is observed in practice is the indirect trace: the fall of serum amyloid A after the peak. A rapid decline suggests that the stimulus has gone, not that a particular resolution pathway has become active.
Acute phase proteins are the practical handle on an invisible process. What they can tell us hangs on their time course, and that differs for every protein. In the horse, white blood cells and serum iron change within hours, serum amyloid A within a day, fibrinogen only after two to three days. From the relation between these three time windows one can estimate how long an inflammatory process has already been running.
| Feature | Horse | Human |
|---|---|---|
| Most important acute phase protein in practice | serum amyloid A | C-reactive protein |
| Role of serum amyloid A | main marker, measurable in the field | biologically highly reactive, rarely measured in routine care |
| Typical baseline in healthy individuals | close to zero | low, but measurable |
| Slower second marker | fibrinogen (two to three days) | erythrocyte sedimentation rate (days) |
| Most common question asked | Is there any inflammation at all, and is it subsiding | How high is the risk, and is the treatment working |
This division of roles is not a biological necessity but the result of measurability and habit. Serum amyloid A is every bit as reactive in the human as in the horse, yet it is rarely measured, because C-reactive protein has established itself as the routine parameter.
A second point is often overlooked: serum amyloid A is not only an indicator, it intervenes in events itself. It binds lipids, influences their transport and acts on other cells in a cytokine-like way. Anyone measuring the value is therefore watching a participant, not a neutral pointer. That also explains why this protein has been conserved across the vertebrates instead of being replaced by something simpler.
The most striking difference is not the kind of response but its gain. In the equine experiment with artificially induced joint inflammation in 24 Standardbred horses, serum amyloid A rose after 36 to 48 hours to about 227 times baseline. Haptoglobin in the same experiment reached only about 1.1 times baseline, while fibrinogen rose on average by about 90 per cent above baseline. Three proteins, the same stimulus, the same animal: more than two orders of magnitude separate the largest and the smallest excursion.
The calibration differs even within the equid family. In a comparison of 47 sick or injured horses, 17 donkeys and 13 mules at one clinic, the values of the mules lay clearly below those of the horses; the authors consider a weaker inflammatory response in mules possible, without being able to demonstrate it, since healthy comparison animals were lacking.
The strongest practical statement concerns early detection. In 122 sport horses flown in for competitions, a serum amyloid A reading taken 24 hours after arrival identified sick animals with about nine out of ten sick animals correctly picked up and just as many healthy animals correctly cleared. Taking the rectal temperature identified only about three out of a hundred sick animals in the same group. The study comes from the Swiss Institute of Equine Medicine in Bern.
In newborn foals the strength lies elsewhere. In a retrospective analysis of 397 foals, septic animals had a median value of 114 micrograms per millilitre against 1.5 in sick but non-septic foals and 0 in healthy ones. At a threshold of 100 micrograms per millilitre the test hardly ever gave a false alarm in healthy and non-septic foals, but it picked up only a good half of the foals that really were septic. A low value therefore does not rule sepsis out.
The negative findings are just as instructive. In 176 adult horses with acute colitis the value on admission did not differ between animals that survived and animals that did not (median values of 548 against 396 milligrams per litre): the marker measures inflammatory activity, not prognosis. Anyone wanting to judge the prognosis had to fall back, in the same study, on lactate, heart rate, age and duration of colic. And in a trial in six healthy horses without a control group, repeated intramuscular injections of procaine penicillin sent several blood values up temporarily in the absence of any disease.
A structured literature review covering pig, horse and cattle accordingly concludes, with due caution, that acute phase proteins currently offer no basis for deciding on the use of antibiotics; any pointer to the underlying cause is at best hinted at in the horse.
In the human the evidence is denser, and for that very reason more sobering. An analysis of 14 studies with 7,755 febrile infants showed that procalcitonin detected invasive bacterial infections considerably better than C-reactive protein; for the wider category of serious bacterial infection both markers were equally middling. A review of 112 papers on bacterial meningitis in childhood found excellent discrimination only for markers in cerebrospinal fluid, and for not a single blood marker.
The most important finding concerns causation. In a genetic analysis of just under 195,000 people from 47 studies, a lifelong inherited elevation of CRP was not associated with more coronary heart disease, even though the observed association was clear in the same data. CRP therefore marks risk; that it causes that risk itself is unlikely on these data.
Consistently enough, there is also little to gain from lowering the marker itself. A systematic review of 15 studies found no relation between the extent of CRP lowering and the extent of risk reduction. A meta-analysis of 14 studies with more than 133,000 people showed that more intensive lipid lowering reduces the event rate without measurably changing CRP: a residue of inflammatory risk remains.
Targeted anti-inflammatory treatment, by contrast, can reduce events. In a randomised trial in 5,522 people with chronic coronary heart disease the composite endpoint occurred in 6.8 per cent on daily low-dose colchicine and in 9.6 per cent on placebo, over a mean follow-up of a good two years. An analysis of 25 randomised trials puts that in perspective: across all anti-inflammatory agents no benefit was demonstrable, but it was for two particular modes of action, and high-sensitivity CRP did not change in the process.
On the nutrition side the effects turn out small and inconsistent. An analysis of 19 randomised trials in overweight found a slight reduction in CRP and tumour necrosis factor alpha on plant-derived alpha-linolenic acid, but at the same time a rise in LDL cholesterol. An analysis of 18 randomised trials with 1,018 people with rheumatoid arthritis found fewer tender joints on omega-3, while the change in CRP and erythrocyte sedimentation rate was not statistically confirmed.
Three statements survive the comparison. First, the architecture of the inflammatory response is built the same way in horse and human: the same recognition pathways, the same signalling molecules, the same switch in the liver. Second, serum amyloid A is highly reactive in both species because it is evolutionarily old. Third, resolution is a process in its own right in both species, and it can fail.
Just as clear is what does not follow. No cut-off, no fold-increase and no timing may be carried over between the species. A high reading yields neither a cause nor a site nor a prognosis. And from the observation that a marker goes together with disease it does not follow that lowering it helps: for CRP precisely this has been tested with genetic data, and the result argues against it.
The marker measures the activity of the fire, not its location and not its cause.
Summarising reading of the cited papers
Original analysis
| Marker | Data basis | First rise | Peak | Size of the excursion | What it does not say on its own |
|---|---|---|---|---|---|
| Serum amyloid A | horse, experimental joint inflammation, 24 animals | from 16 hours | 36 to 48 hours | about 227 times baseline | whether the cause is infectious or non-infectious |
| Haptoglobin | horse, same experiment | from 24 hours | 48 to 96 hours | about 1.1 times baseline | reacts too weakly for the early phase |
| Fibrinogen | horse, same experiment | from 24 hours | 36 to 72 hours | rise of about 90 per cent above baseline | is also moved by clotting processes |
| White blood cells and serum iron | horse, clinical review | within hours | hours to one day | no consistent figure given | not very specific, strongly influenced by stress |
| Serum amyloid A | vertebrates in general, review | hours | about 24 hours | up to about 1000-fold | is hardly ever measured in human routine care |
| C-reactive protein | human, meta-analysis in febrile infants | hours | one to two days | thresholds between 13 and 20 milligrams per litre discussed | separates invasive bacterial infection only moderately |
No. Serum amyloid A rises with any sufficiently strong inflammation, whatever the cause. In one experiment a purely chemically induced joint inflammation, with no pathogen at all, triggered a rise to about 227 times baseline. Repeated intramuscular injections can also lift the value temporarily in healthy horses, and reviews describe rises after surgery. The marker answers the question of whether an inflammation is running and how strongly. Whether bacteria are involved, whether treatment is needed and where the focus lies has to be settled by the clinical examination.
Because different markers have established themselves in practice in the two species. In the horse, serum amyloid A reacts quickly and strongly and can be measured stable-side, whereas fibrinogen takes two to three days. In the human, C-reactive protein has been the routine parameter for decades and carries the largest body of data. Biologically, serum amyloid A is highly reactive in the human too, it is simply rarely measured. The choice of marker therefore reflects measurement habit and availability, not a fundamental difference in the biology of inflammation.
That conclusion is not carried by the data. In a genetic analysis of just under 195,000 people, those with an inherited lifelong elevation of CRP did not have more coronary heart disease, even though the observed association was clear. A systematic review also found no relation between the extent of CRP lowering and the extent of risk reduction. Effective anti-inflammatory treatment acts on particular signalling pathways, not on the reading. This is not a treatment recommendation; decisions about therapy belong in a doctor's hands.
A rapid fall suggests that the inflammatory stimulus is no longer acting. That is the clinically most useful property of this marker: it rises quickly and falls quickly as soon as the inflammation subsides, and so it makes the course visible. It is not, however, a prognostic value. In a study in 176 horses with acute colitis the value on admission did not differ between animals that survived and animals that did not. Assessment and further steps belong in a vet's hands.
The expectation is bigger than the evidence. Omega-3 fatty acids do indeed give rise to the body's own resolution mediators, and that is well established mechanistically. In clinical trials in humans, however, the effects on inflammatory markers remain small: in rheumatoid arthritis, an analysis of 18 studies showed an improvement in tender joints, while the change in CRP and erythrocyte sedimentation rate was not statistically confirmed. For the horse, comparably robust studies are missing altogether.
Because the same signalling molecules that attract immune cells also make the pain fibres in the tissue more sensitive. That coupling is functional: pain forces the affected area to be spared and so protects the healing. In the horse it shows as lameness, in the human as pain on loading. In the equine experiment mentioned above, the lameness began to decrease after 24 to 36 hours, while the blood marker reached its peak only afterwards: complaints and reading do not run in step.
Terms defined in this text
Comparative Health
Radiocarbon data suggest the Achilles tendon core forms in growth and is barely renewed thereafter: what that means for healing, scar and the equine t
MUMultiple speciesBmoderateT2limited target data Open access
Equine Health
What counts is not visible dust but the respirable fraction in the horse's breathing zone: figures from trials on hay, steaming and haylage.
MUMultiple speciesBmoderateT1hypothesis only Open access
Evidence Lab
Evidence levels grade confidence, not truth: what randomised, observational and animal studies support, how GRADE rates them and where the pyramid bre
MUMultiple speciesBmoderateT0not transferable Open access
Once a month: new syntheses, checked studies, and what has changed in the state of the evidence.