asymptomatic or associated with adverse hemodynamic and metabolic consequences. Specifically, elderly patients, often those taking cox-2 inhibitors, have renal insufficiency and can develop a serious metabolic acidosis. The negative cardiac inotropic actions of acidosis are well characterized. It seems intuitive that the manufacturers of these 3 cox-2 preferential inhibitors would have examined the molecular structures of these drugs long before marketing them and would have characterized the predictable side effects of celecoxib and valdecoxib knowing the relevant pharma- cophore, the unsubstituted sulfonamide. A second observation that intrigued us occurred during our review of the Product Information sections in the 2000 to 2005 editions of the Physicians’ Desk Reference (PDR; Montvale, NJ: Medical Economics Company). 5 We viewed these data to determine whether or not adverse events known to be associated with carbonic anhydrase inhibitors were listed for the cox-2 inhibitors. We also examined rofecoxib, even though it is not a carbonic anhydrase inhib- itor. The addition of an adverse event to the Product Infor- mation section of the PDR represents the conclusion of a process of signal detection and evaluation. We report that the Laboratory Tests subsection listed under Precautions in the 2000 edition of the PDR states that, during controlled clinical trials, there was an increased incidence of hyper- chloremia in patients receiving celecoxib compared with patients receiving placebo. Exposure time to the drug was probably less than 6 months. Other laboratory abnormalities that occurred more frequently in patients exposed to cele- coxib included hypophosphatemia and elevated blood urea nitrogen. There was no mention of serum bicarbonate values; nevertheless, a serum bicarbonate value may be approxi- mated. If the anion gap (when potassium is not included) is 6 to 13 mM (average 8 mM) and the serum sodium value is 140 mmol/L, then an abnormally elevated serum chloride value of 114 mmol/L would correspond to a bicarbonate value of 18 mmol/L (the common reference value for ve- nous plasma or serum is 22-30 mmol/L). Some patients treated with celecoxib (duration unknown) were reported to have peak serum chloride values at last visit of 115 mmol/L and higher. There was a trend toward patients having ab- normalities in both serum chloride and phosphate levels. Therefore, we deduced that hyperchloremic, nonanion gap metabolic acidosis may in some patients be associated with exposure to celecoxib. In fact, it is known that certain patients given cox-2 inhibitors over prolonged periods de- velop a form of low renin-angiotensin activity characterized by a modest hyperkalemia and a mild metabolic acidosis. The laboratory findings of hyperchloremia and hy- pophosphatemia were not listed in the PDR in the years following 2002. Renal calculus, another side effect associ- ated with carbonic anhydrase inhibition, was mentioned for celecoxib in the Adverse Reactions section of the PDRs in all years. The Product Information sections for valdecoxib (PDR 2002-2005 editions) and rofecoxib (PDR 2000-2005 editions) do not mention dysfunction of renal acidification such as metabolic acidosis, hyperchloremia, hypophos- phatemia, and defective bicarbonate reclamation. Renal cal- culus and urolithiasis were, however, reported in the Ad- verse Reactions sections of PDRs for valdecoxib and rofecoxib, respectively. Our results have implications in several areas of interest, not the least of which are the apparent multifaceted mech- anisms of action of celecoxib. Based on our findings, the cox-2 inhibitor celecoxib, unlike rofecoxib, has another mechanism of action, that of inhibiting carbonic anhydrase II. Acute exposure to celecoxib initially may, in a fashion analogous to that of acetazolamide, induce a mild diuresis, unlike rofecoxib, which only inhibits prostaglandin synthe- sis. However, this sparing effect of the diuresis is self- limited. The continued exposure to celecoxib not only ex- poses a patient to the structure-adverse events of cox-2 inhibition, that of blocking endothelial prostacyclin produc- tion and the theoretical arterial thrombosis, but also to the structure-adverse events associated with the pharmacophore -SO 2 NH 2 and carbonic anhydrase inhibition. James F. Knudsen, MS, PhD, MD New Hope Cancer Center Hudson, Fla Gerald H. Sokol, MS, MD H. Lee Moffitt Cancer Center at the University of South Florida College of Medicine Tampa, Fla doi:10.1016/j.amjmed.2005.06.069 References 1. Alpert JS. The Vioxx debacle. Am J Med. 2005;118:203. 2. Knudsen J, Sokol GH, Cantilena LR. Structure-activity relationship as predictors of adverse drug events (ADEs). Clin Pharmacol Ther. 2003; 73:39-40. 3. Knudsen JF, Carlsson U, Hammarström P, Sokol GH, Cantilena LR. Cox-2 inhibitors and carbonic anhydrase activity [abstract]. Clin Phar- macol Ther. 2004;75:44. 4. Weber A, Casini A, Heine A, et al. Unexpected nanomolar inhibition of carbonic anhydrase by COX-2-selective celecoxib: new pharmacologi- cal opportunities due to related binding site recognition. J Med Chem. 2004;47:550-557. 5. Physician’s Desk Reference. Montvale, NJ: Thomson PDR. Mechanical Ventilation Learns New Tricks To the Editor: The rapidly evolving field of noninvasive positive pres- sure ventilation (NPPV) makes necessary the inclusion of two additional studies in the excellent review by Calfee and Matthay on advances of mechanical ventilation. 1 First, Diaz et al performed a prospective, open, noncon- trolled study to assess the outcomes of NPPV therapy in patients with a Glasgow Coma Scale score of 8 due to acute respiratory failure and found that a response rate of 80% (76/95) to NPPV therapy compared with a response 94 The American Journal of Medicine, Vol 119, No 1, January 2006