NSAID's: Comparative toxicity and drug interactions (Proceedings)

With the availability of many over-the-counter NSAID's, and several newer, relatively cyclooxygenase-2 (COX-2) selective agents marketed for veterinary patients, there are many choices for controlling acute and chronic pain and inflammation. Because patients are often given NSAID's in the presence of concurrent disease and along with other drugs, the potential for NSAID-related organ toxicity and NSAID-associated drug interactions should always be considered.

Differential toxicity of available NSAID's

      GI ulceration

All NSAID's (even those marketed as COX-2 selective) have the potential to cause GI bleeding, ranging from mild mucosal erosions to full thickness ulcers and perforation. NSAID's impair gastric mucosal integrity by inhibition of the synthesis of prostaglandins, which are needed to increase epithelial cell turnover, and mucus and bicarbonate secretion. NSAID's also directly alter phospholipids in the mucus gel layer overlying the gastric mucosa, and can damage this hydrophobic barrier (Lichtenberger 1995). The ulcerogenic effects of NSAID's are potentiated by multiple NSAID use (Reed 2002), and by glucocorticoids (Boston 2003) via inhibition of prostaglandin synthase and decreased peroxidase-mediated scavenging of free radical precursors (Bandyopadhyay 1999). Therefore, the use of multiple NSAID's, or NSAID's with glucocorticoids (or, presumably, in dogs with untreated hyperadrenocorticism), is contraindicated.

It is difficult to compare with confidence the relative GI toxicity of available NSAID's, since most published in vivo studies that use endoscopy (the gold standard) enroll relatively few dogs. In one placebo-controlled endoscopic study comparing 4 weeks of daily meloxicam, carprofen, or ketoprofen at label dosages in dogs, carprofen was associated with the fewest and mildest lesions, although no drug was associated with clinical signs of GI bleeding (Forsyth 1998). In another multi-dose study, carprofen and etodolac were both well tolerated when endoscopic lesions were compared to those seen with buffered aspirin (Reimer 1999).

Anti-inflammatory agents in the coxib family, which are selective for inhibition of cyclooxygeanse-2 (COX-2), include the veterinary products deracoxib (Deramaxx), and firocoxib (Previcox), and human products such as celecoxib (Celebrex) and rofecoxib (Vioxx). These drugs have a better safety profile than classical NSAID's in humans, and overall may carry a lower risk of GI bleeding in dogs. However, gastrointestinal ulceration and perforation have been reported in dogs with deracoxib, and this has now been added to the Deramaxx product label. Because of the potential risk of GI bleeding with any NSAID, client education and careful monitoring for GI upset, darkened stools, inappetance, or lethargy should be instituted in any treated patient.

Because life threatening GI bleeding can occur with minimal prodromal signs, periodic monitoring of CBC's for anemia, decreased total protein, polychromasia, or microcytosis is advised in older patients or in those with underlying risk factors for GI bleeding, such as hepatic disease or early renal insufficiency.

Gastric acid is necessary for the development of gastric ulcers associated with NSAID's (Scarpignato 1999). Misoprostol reduces ulcer incidence and ulcer complications from NSAID's in humans, but misoprostol can cause diarrhea and cramping. Proton pump inhibitors or high doses of H2 blockers are also effective in reducing the incidence of both gastric and duodenal ulceration from NSAID's, without causing diarrhea (Rostom 2002). In dogs, misoprostol reduces gastroduodenal ulceration and vomiting from aspirin (Murtaugh 1993), and q. 12 h dosing appears to be as effective as q. 8 h. dosing (Ward 2003). In dogs, the effect of H2 blockers or pump blockers on NSAID-induced ulcers remains to be convincingly demonstrated (Jenkins 1991); however, omeprazole is a reasonable empirical choice for NSAID ulcer prophylaxis in dogs.

      Renal decompensation

All NSAID's can adversely affect renal perfusion, since PGE2 and PGI2 mediate afferent arteriolar dilation in response to decreased renal blood flow, and PGI2 stimulates renin release (review Forrester 1999). The risk of renal decompensation from NSAID's is greatest in human patients with pre-existing renal disease, hypovolemia/dehydration, congestive heart failure, sodium-restricted diets, or cirrhosis. All of these states are dependent on elaboration of renal prostaglandins to maintain renal blood flow.

COX-2 is constituitively expressed in the kidney, so even COX-2 selective agents can decrease glomerular filtration to the same extent as classical NSAID's. Most studies in dogs have been done in healthy patients undergoing elective procedures, and under such conditions, carprofen at the label dose has been shown to have no adverse effect on glomerular filtration rate (GFR), as determined by scintigraphy (Ko 2000; Bostrom 2002), although another study showed a decrease in endogenous creatinine clearance postoperatively in healthy dogs undergoing castration, following treatment with either carprofen or ketoprofen (Forsyth 2000). In addition, ketoprofen decreases natriuresis postoperatively and can be associated with transient azotemia, even in healthy dogs undergoing spays (Lobetti 2000).

      Platelet dysfunction and bleeding

Coxibs such as celecoxib, and rofecoxib do not inhibit platelet function in humans. Similarly, deracoxib and firocoxib do not prolong buccal mucosal bleeding time in dogs. However, less selective NSAID's may inhibit platelet function by inhibiting of COX-1-mediated TXA2 generation.

Etodolac has been associated with excessive bleeding in dogs during experimental surgery (Etodolac label). Carprofen has been associated with mild subclinical decreases in platelet aggregation in vitro (Hickford 2001), although neither carprofen nor meloxicam are associated with prolongation of buccal mucosal bleeding times in healthy dogs (Hickford 2001, Mathews 2001). In one randomized prospective study of 93 dogs undergoing elective orthopedic surgery (Grisneaux 1999), bleeding times were prolonged in dogs treated with ketoprofen compared to those given either carprofen or placebo, and one ketoprofen-treated dog developed a hematoma at the surgical site. Therefore, in dogs with pre-existing coagulopathies, such as von Willebrand's disease, carprofen, meloxicam, deracoxib or firocoxib may be better NSAID choices than less COX-2 selective drugs.

      Hepatopathy

Carprofen has been associated with rare acute hepatic necrosis in dogs, with most dogs affected 14 to 30 days after drug initiation (MacPhail 1998). The mechanisms for this idiosyncratic reaction are not known. Although Labrador retrievers were over-represented in the initial report, Pfizer has been unable to reproduce the syndrome in Labradors. The incidence of any reported hepatopathy associated with carprofen is approximately 0.05% (~5 cases per 10,000 dogs treated).

Icterus and hepatic enzyme elevations have also been reported in association with deracoxib administration in dogs, although the incidence is not yet clear. In humans, cholestatic hepatopathy has been reported as an idiosyncratic reaction to either celecoxib or rofecoxib, and fulminant hepatic failure has been associated with etodolac. It appears that most NSAID's have the potential to cause hepatopathy. Careful clinical monitoring, as well as evaluation of ALT, SAP, bilirubin, and albumin in any dog showing even mild clinical deterioration during NSAID administration, is important.

      Thyroid dysfunction

Etodolac (Etogesic) administration has been associated with a fall in serum total T4 in dogs with orthopedic disease, sometimes into the hypothyroid range (Ness 2003). While this has been attributed to protein binding displacement of thyroxine by NSAID's (leading to increased free T4, decreased TSH, and feedback inhibition of T4 synthesis), no increases in free T4 and/or decreases in TSH have been documented. In addition, another study found no changes in thyroid parameters in healthy dogs given etodolac (Panciera 2002).

Meloxicam and carprofen have no effect on T4, free T4, or TSH when given at the label doses over 60 days (Sauve 2003). It is possible that NSAID administration may cause a decrease in total T4 in some dogs, although the effect of concurrent illness should also be considered. Dogs treated with NSAID's and suspected of hypothyroidism should be evaluated with a full thyroid panel, to include free T4 and canine TSH.

Drug Interactions with NSAID's

      Protein binding

Most NSAID's, including coxibs, are highly protein bound, and could potentially lead to displacement and increased toxicity from other highly protein bound drugs. In this context, protein binding is only clinically relevant if it is >90%. For example, for a compound that is 99% protein bound, 1% of plasma drug is free and active. A second drug may cause only a small decrease in protein binding of the first drug (e.g. 99% to 96%), but the amount of free drug quadruples from 1% to 4%, as if you had given four times the dose. For a drug that is only 55% protein bound, a second drug may again cause a small decrease in protein binding of the first drug (e.g. 55% to 52%), but this time the amount of free drug increases from 45% to 48% (a much smaller relative increase, and clinically insignificant).

Highly protein bound drugs that interact with NSAID's in humans include warfarin, valproic acid, methotrexate, and diazepam. Drug dose adjustments for protein binding are difficult to make clinically because often protein binding data are not available for dogs and cats. There can be marked species differences in protein binding, so information from human drug inserts may not be accurate for veterinary patients. Most importantly, increases in free drug that result from protein binding interactions are offset over time by increased elimination of the extra free drug.

It is probably reasonable to consider dose reductions when giving single or loading doses of relatively toxic drugs that are > 90% protein bound, to animals that have low serum albumin concentrations or are being treated with other highly protein bound drugs. This situation would apply to NSAID's (or benzodiazepines) being given in the perioperative period, but not for NSAID's given chronically.

      Hemodynamic effect

NSAID's may complicate the management of congestive heart failure, since NSAID's may increase blood pressure (due to loss of vasodilatory prostaglandins), and increase salt and water retention (due to loss of renal prostaglandins that promote natriuresis). In humans, NSAID's can blunt the anti-hypertensive effects of ACE inhibitors and beta blockers. In addition, NSAID's can blunt the diuretic response to furosemide and thiazides, and may exacerbate hyperkalemia when used with potassium sparing diuretics (Webster 1985).

These effects vary with the type of NSAID, and the dose and duration of therapy. For example, indomethacin and phenylbutazone are implicated most commonly in humans, while meloxicam does not significantly affect furosemide action in human (Muller 1997). Even coxib agents may counteract the efficacy of furosemide or antihypertensive drugs. NSAID's increase the nephrotoxicity of aminoglycosides, via both decreased clearance of aminoglycosides and inhibition of compensatory prostaglandin production.

      Absorption and hepatic metabolism

Acetaminophen is metabolized to its toxic reactive intermediate, NAPQI, by cytochrome P4502E1 in humans. This pathway is induced by phenobarbital, rifampin, and ethanol. Chronic phenobarbital administration increases the hepatotoxicity of acetaminophen in experimental studies, as does alcoholism in humans. Metoclopramide increases gastric emptying of acetaminophen and aspirin, and can enhance intestinal absorption, and toxicity, in cases of acetaminophen or aspirin overdose in humans.

      Interactions of NSAID's with herbal supplements

Owners of chronically ill pets may add herbal supplements to their prescribed drug regimen, in hopes of improving therapeutic response. However, several popular over-the-counter herbal supplements may the potential for have adverse interactions with NSAID's. For example, gingko, garlic, ginger, and ginseng inhibit platelet aggregation, and gingko has been reported in humans to lead to spontaneous bleeding in combination with aspirin. Herbs that contain salicylate, such as meadowsweet and willow, could also exacerbate the side effects of aspirin and other NSAID's.