Flow diagram of study selection process. CIN indicates contrast-induced nephropathy; RCTs, randomized controlled trials.
Forest plot of odds ratios for developing contrast-induced nephropathy from 6 trials. CI indicates confidence interval.
Begg funnel plot with pseudo 95% confidence intervals to assess for evidence of publication bias.
Forest plot of differences in serum creatinine levels between theophylline and control groups at 48 hours after contrast media administration from 7 trials. The error bars represent 95% confidence intervals.
Bagshaw SM, Ghali WA. Theophylline for Prevention of Contrast-Induced NephropathyA Systematic Review and Meta-analysis. Arch Intern Med. 2005;165(10):1087-1093. doi:10.1001/archinte.165.10.1087
Contrast-induced nephropathy (CIN) is an important cause of declines in kidney function and is related to greater morbidity, health care costs, and mortality. Adenosine has been proposed to contribute to the pathophysiological process of CIN. We performed a systematic review and meta-analysis of theophylline, an adenosine antagonist, for the prevention of CIN.
Studies were identified in all languages by search of MEDLINE (1966 through November 2003), EMBASE (1980 through week 44 [November] of 2003), and the Cochrane Controlled Clinical Trials Register (1996 through November 2003) databases and selected conference proceedings.
We searched for randomized controlled trials comparing theophylline vs control in patients receiving radiocontrast media for angiography or computed tomography.
Our primary outcome measures were the risk of CIN, the difference in serum creatinine levels between theophylline and control groups at 48 hours and need for dialysis.
Nine randomized controlled trials involving 585 patients were identified and included for analysis. Theophylline protocols and definitions of CIN varied across studies. There was evidence of heterogeneity of results across trials (Q = 9.77; P = .08); therefore, pooled values require cautious interpretation. The overall pooled odds ratio (OR) using a conservative random-effects model was 0.40 (95% confidence interval [CI], 0.14 to 1.16; P = .09) indicating a trend toward reduction in the incidence of CIN with theophylline use. The pooled estimate for the difference in 48-hour serum creatinine levels between the theophylline and control groups was –0.17 mg/dL (95% CI, –0.28 to –0.06 mg/dL) (−15.2 μmol/L [95% CI, −24.6 to −5.7 μmol/L]) (P = .002), indicating that theophylline may be protective in CIN. The incidence of CIN requiring dialysis was uncommon and reported in only 1 case.
Theophylline may reduce the incidence of CIN with an efficacy that is perhaps comparable to that reported in studies of N-acetylcysteine. However, findings are inconsistent across studies. A large, well-designed trial that incorporates the evaluation of clinically relevant outcomes is required to more adequately assess the role for theophylline in CIN prevention.
Contrast-induced nephropathy (CIN) is an important cause for iatrogenic acute renal failure1,2 and is associated with incident dialysis, prolonged hospitalization, increased health care costs, potentially irreversible reductions in kidney function, and mortality.3- 7 Although incompletely understood, the pathophysiological process of CIN is hypothesized to occur via contrast media–induced vasoconstriction and medullary ischemia coupled with the generation of free radicals and oxidative injury to tubular cells.8- 11
There is evidence demonstrating that an elevated endogenous adenosine level may contribute to the pathophysiological process of acute reductions in kidney function following radiocontrast media exposure.12,13 Increased urinary excretion of adenosine has been demonstrated following the intravascular administration of radiocontrast media.14 Adenosine can induce sustained renal vasoconstriction and a reduction in glomerular filtration rate.
Adenosine receptor antagonists may attenuate the vasoconstrictive effects observed with radiocontrast media and preserve both renal blood flow and glomerular filtration perfusion pressure.15,16 Several small studies have assessed the efficacy of theophylline for the prevention of CIN.17- 26
The specific objectives of this meta-analysis were to assess the effect of theophylline on (1) the dichotomous end point of CIN (yes/no) and (2) serum creatinine levels following the administration of contrast media. We also conducted a meta-regression analysis to determine whether particular clinical or study quality factors influence the apparent effect of theophylline on risk of CIN.
We identified published randomized controlled trials (RCTs) of adenosine antagonists for the prevention of CIN during intravascular angiography and computed tomography using both electronic and manual search strategies. We supplemented this by scanning the reference lists of all identified articles, reviewing conference proceedings of 3 large international meetings (American Heart Association, American Society of Nephrology, and American College of Cardiology) and by contacting experts in the field. All languages and types of publications were considered eligible. The comprehensive literature search was performed in November 2003 and updated in June 2004 with a verification search for any new studies.
MEDLINE (1966 through November 2003), EMBASE (1980 through week 44 [November] of 2003), and the Cochrane Controlled Clinical Trials Register (1996 through November 2003) databases were searched via OVID using an approach recommended for systematic reviews of randomized trials.27 PubMed was also searched.28 We derived 3 comprehensive search themes that were then combined using the Boolean operator “and.” The first theme used a recommended highly sensitive RCT filter and systematic review filter method.29 The second theme, “contrast-induced nephropathy,” was created by using the Boolean search term “or” to search for the following terms appearing as both exploded MeSH headings and text words: contrast media or contrast or radiocontrast or kidney or kidney failure or acute renal failure or nephropathy. The third theme, “adenosine antagonists,” was created by a search using an exploded MeSH heading and text word search for the terms theophylline or aminophylline.
We independently evaluated identified articles for eligibility on the basis of 4 inclusion criteria: (1) study design (RCTs), (2) target population (patients receiving contrast media for intravascular angiography or computed tomography), (3) intervention (trials of adenosine antagonists vs control), and (4) outcome (a priori explicit definition of CIN or use of change in serum creatinine level or glomerular filtration rate prior to and following contrast administration for the primary outcome).
We independently extracted data from all primary studies fulfilling eligibility criteria. Any discrepancies in extracted data were resolved by consensus. Data extracted included identifying information, focus of the study, details of study protocol, and demographic data. The primary outcome measures were the incidence of CIN, change in serum creatinine levels, and change in glomerular filtration rate. The secondary outcome measures were requirement of renal replacement therapy and adverse reactions to adenosine antagonists. Authors of the studies were contacted for additional information when applicable.
We independently assessed the methodological quality of individual studies. Any disagreements were resolved by consensus. Items used to assess study quality were methods of randomization, any blinding, use of a placebo, reporting of losses to follow-up or missing outcome assessments, and evidence of important baseline differences between the groups.30- 32 An overall quality score was determined for each study as described by Jadad et al.30
Data from all of the selected RCTs were combined to estimate the pooled odds ratio (OR) with 95% confidence intervals (CIs) using a random-effects model as described by DerSimonian and Laird33 and Senn et al.34 The presence of heterogeneity across trials was evaluated using a χ2 test for homogeneity.35 Because the χ2 test has a low sensitivity for detecting heterogeneity, a P value of ≤.10 was considered significant for the presence of statistical heterogeneity.36 Meta-regression was performed to analyze for potential clinical and study quality factors that may influence treatment effects. We tested for potential publication bias using both the Begg37 test for funnel plot asymmetry and the Egger test.38 An additional sensitivity analysis was performed to determine the impact that the addition of a varying number of hypothetical negative trials would have on the pooled results. All statistical analyses were performed with Stata version 8.0 (StataCorp, College Station, Tex).
A total of 57 unique citations were identified by our search strategy (Figure 1). After the initial screen, 26 citations warranted further review. Among these, 17 articles were excluded. Overall, 9 studies were identified and fulfilled our inclusion criteria.17- 25 All of these citations were identified by the electronic search strategy. There was excellent overall agreement for inclusion of individual studies (κ = 0.74).
The RCTs were published between 1992 and 2003. Table 1 and Table 2 present the characteristics and quality indicators of the 9 trials. A total of 585 patients were examined in these 9 trials, among whom 295 received adenosine antagonists and 290 were in control groups. There were 257 patients (44%) with diabetes mellitus, of whom 123 were assigned to receive adenosine antagonists and 134 were assigned to a control group. One study excluded patients with diabetes mellitus20 and another did not report the proportion of diabetic patients.17 Four studies had a priori hydration protocols,19,21- 23 2 studies advised intake of greater than 2 L of fluid prior to contrast media exposure,24,25 and 3 did not detail whether hydration was provided.17,18,20 The type of contrast media used varied across studies. Patients either received ionic or high osmolar,20,21,23 nonionic or low osmolar,18,22,24,25 or both17,19 contrast media
The definitions of CIN and protocols for administration and total cumulative dosages of theophylline varied across studies (Table 1). Key elements of study methods and reporting also varied across studies (Table 2).
The incidence of CIN varied across studies reporting primary dichotomous outcomes. Table 3 and Figure 2 present information on the incidence of CIN for available studies. Three studies provided evidence of a risk reduction for developing CIN with theophylline use,23- 25 whereas 3 studies did not find a statistically significant reduction in risk associated with theophylline treatment.17,21,22 The overall pooled OR for developing CIN using a random-effects model was 0.40 (95% CI, 0.14-1.16; P = .09), suggesting a trend toward a reduced incidence of CIN with theophylline use that is not statistically significant in the more conservative random-effects analysis (Figure 2). However, the pooled effect estimate comparing the occurrence of CIN across all studies needs to be viewed with caution given the evidence of statistical heterogeneity of results across studies (χ2 = 9.77; P = .08).
There was no evidence to suggest publication bias according to both the Begg test (P = .85) and Egger test (coefficient, 0.81; 95% CI, −7.52 to 5.90; P = .75). Figure 3 demonstrates this by the relative symmetry in the Begg funnel plot. The occurrence of CIN requiring dialysis was uncommon and reported in only 1 case.
Table 3 gives a summary of the changes in serum creatinine level available across all studies. The pooled estimate (using a random-effects model) for the difference in 48-hour serum creatinine level between the theophylline and control groups was –0.17 mg/dL (95% CI, –0.28 to –0.06 mg/dL) (–15.2 μmol/L [95% CI, –24.6 to –5.7 μmol/L]) (P = .002) based on data available from 7 studies (Figure 4). This pooled estimate also requires cautious interpretation due to availability of data from only 7 studies and given evidence of statistical heterogeneity of creatinine results across studies (Q = 21.7; P<.001).
Meta-regression was performed to assess a number of clinical and study quality factors that may have contributed to heterogeneity across studies. These analyses suggest that the heterogeneity across studies may be partially explained by the year of study publication (coefficient, –0.28; 95% CI, –0.53 to –0.03; P = .03), with more recent year of publication being associated with a more significant reduction in the incidence of CIN with theophylline use.
Although the heterogeneity was not accounted for by differences in specific study quality factors, including use of placebo for blinding (coefficient, –0.15; 95% CI, –2.58 to 2.28; P = .90), use of any blinding (coefficient, –0.41; 95% CI, –1.57 to 0.75; P = .49), or consecutive patient enrollment (coefficient, –1.67; 95% CI, –3.58 to 0.23; P = .09), the overall Jadad study quality score may partially account for the heterogeneity across trials (coefficient, 1.31; 95% CI, 0.25 to 2.37; P = .02). When the analysis was restricted to the 4 studies with overall Jadad scores of 2 or higher, the overall pooled OR for CIN using a random-effects model was 0.77 (95% CI, 0.29 to 2.04; P = .60), whereas for studies with an overall Jadad scores of lower than 2, the pooled OR was 0.12 (95% CI, 0.03 to 0.42; P<.001), suggesting that studies with lower quality report greater reductions in the incidence of CIN with theophylline use.
Other meta-regression analyses demonstrated that the heterogeneity could not be accounted for by differences in patient age (coefficient, 0.11; 95% CI, –0.04 to 0.27; P = .16), baseline serum creatinine levels (coefficient, 0.03; 95% CI –0.01 to 0.07; P = .16), diabetes mellitus (coefficient, –0.02; 95% CI, –0.06 to 0.03; P = .52), whether the procedure was performed electively or emergently (coefficient, 2.19; 95% CI, –0.51 to 4.89; P = .11), or periprocedure hydration administered (coefficient, 1.08; 95% CI, –1.17 to 3.33; P = .35). In addition, the heterogeneity was not accounted for by differences in the total dose of adenosine antagonist administered (coefficient, 0.001; 95% CI, –0.0002 to 0.001; P = .13) or the volume of contrast media administered (coefficient, 0.004; 95% CI, –0.02 to 0.03; P = .71). There was a trend for reduction in CIN when ionic vs nonionic contrast media was used (coefficient, 2.14; 95% CI, –0.59 to 4.86; P = .13). This was further supported in the study by Kapoor and colleagues23 in which ionic radiocontrast media was used. That study demonstrated the largest reduction in incidence of CIN with an estimated OR of 0.06 (95% CI, 0.01 to 0.53), a finding that suggests that theophylline may be more protective for patients receiving ionic as opposed to nonionic contrast media.23
To explore the sensitivity of our final results to the existence of potential unpublished “negative” trials, we modeled the existence of a varying number of hypothetical negative trials, each enrolling 100 patients, with an estimated event rate of CIN of 19% in each control group and yielding an OR of 1.2. Using a random-effects model, the addition of 7 such trials would result in a pooled OR of 1.01 (95% CI, 0.73 to 1.39; P = .90). If fewer than 7 such trials exist, the pooled OR point estimate remains below 1.0. These results imply that the pooled OR results suggesting benefit with theophylline use are relatively robust to the addition of new negative trials.
This systematic review combines results from 9 randomized studies assessing the efficacy of adenosine antagonists (theophylline or aminophylline) for the prevention of CIN and acute elevations in serum creatinine level in patients administered radiocontrast media. There is a pathophysiological rationale for the prophylactic administration of adenosine-receptor antagonists to potentially attenuate the renal toxic effects observed with radiocontrast media. Our results indicate that the use of adenosine antagonists is associated with a trend toward reduced incidence of CIN and a small reduced pooled mean difference in serum creatinine level at 48 hours.
The pooled results that we report need to be interpreted with some caution, given the evidence of significant heterogeneity of results across studies. Given this heterogeneity, we have elected to present the more conservative random-effects pooled estimate for dichotomous outcome of CIN, and the resulting pooled OR has a 95% CI that surpasses 1.0.
Meta-regression analysis revealed 3 factors that may partially contribute toward this trend for heterogeneity. First, differences in study quality can represent a potentially important source of heterogeneity.39 In general, the trial quality scores were uniformly low. We found that studies with lower overall quality scores were associated with a greater reduction in risk for CIN. Similarly, studies that were more recently published were associated with a greater reduction in incidence of CIN. Second, ionic radiocontrast media is associated with an increased incidence of CIN in patients with reduced baseline kidney function.40 We demonstrated a trend for a reduced incidence of CIN with theophylline administered prior to exposure with ionic radiocontrast media (as opposed to nonionic contrast). This raises the possibility that theophylline may be more protective for patients receiving more “toxic” ionic contrast.23 Finally, variation in the theophylline protocols (ie, total dose and schedule) across studies may also represent a source of heterogeneity. We found that theophylline prophylaxis was associated with a larger overall benefit for studies with no predefined hydration protocol, suggesting that theophylline may render some additional benefit in patients not receiving any hydration. In a test of this hypothesis, Bader and colleagues41 administered theophylline prior to radiocontrast media exposure to 19 patients with congestive heart failure and reduced baseline kidney function unable to receive hydration. The overall incidence of CIN was 21%, with no patient requiring acute dialysis in this cohort. Although this trial had no control group, the authors suggested a preservation of glomerular filtration and avoidance of dialysis with the use of theophylline. Similarly, Bader and colleagues42 compared prophylactic theophylline to N-acetylcysteine in patients with reduced baseline kidney function unable to receive hydration prior to radiocontrast media exposure. A total of 36 patients were randomized to receive either theophylline or N-acetylcysteine intravenously prior to the procedure. Theophylline was superior, with a greater reduction in the incidence of CIN compared with N-acetylcysteine (28% vs 39%; P = .045). At 1 week of follow-up, no patient required rescue dialysis in the theophylline group compared with 2 in the N-acetylcysteine group.
Few studies reported whether any significant adverse effects occurred with use of theophylline. Huber and colleagues24 reported a temporary increase in heart rate (<15/min) immediately following administration of theophylline, 200 mg intravenously, without significant adverse sequelae. However, use of theophylline in select populations may be limited by relative contraindications, particularly patients with active coronary ischemia, arrhythmias, or preexisting seizure disorders. Only 2 studies measured serum theophylline levels. Gandhi and colleagues17 reported a mean serum theophylline level of 5.8 μg/mL (32.2 μmol/L) immediately prior to angiography after receiving three 125-mg doses in the proceeding 24 hours. Erley and colleagues18 administered theophylline, 5 mg/kg (mean weight of patients, 74 kg), 45 minutes prior to radiocontrast media exposure and reported a mean serum theophylline level of 7.2 μg/mL (40.1 μmol/L) 1 hour after the procedure. To our knowledge, no studies to date have assessed whether serum theophylline level corresponds to a dose-response reduction in the incidence of CIN. Despite the relative low dose of theophylline administered in these trials, any future studies should incorporate potential adverse effects as an important secondary outcome.
To our knowledge, this is the first study to assess the pooled evidence for prophylactic use of theophylline for the prevention of CIN. This meta-analysis suggests that theophylline may reduce the incidence of CIN; however, the currently available evidence remains inconclusive and warrants renewed interest to establish whether patients may derive a clinical benefit. Our sensitivity analysis of hypothetical negative trials suggests that many such trials would be required to neutralize the trend toward a potential protective effect of theophylline. Despite this, however, we emphasize that there is considerable heterogeneity of effect across studies that raises some question of the true efficacy of theophylline for the prevention of CIN.
Of note, these results appear analogous to the current body of evidence on using prophylactic N-acetylcysteine for the prevention of CIN. Despite generally promising results to support the efficacy of N-acetylcysteine, several clinical trials have yielded conflicting results on the efficacy of N-acetylcysteine, while a series of completed meta-analyses speculated that although N-acetylcysteine may be of benefit, its efficacy remains inconclusive owing to the heterogeneity across studies and evidence of publication bias.43- 47 Perhaps a well-designed, RCT comparing N-acetylcysteine, theophylline, and N-acetylcysteine plus theophylline vs placebo may further provide important insight into the comparable efficacy of these regimens for the prevention of CIN.
Our findings indicate that the evidence supporting the use of theophylline for the prevention of CIN, though promising, remains inconclusive. The results of the trials that we reviewed to date should be viewed as being collectively suggestive of possible benefit; however, a larger, more definitive, well-designed trial for confirmation and assessment of risks in certain populations is needed before theophylline could be recommended routinely for the prevention of CIN. Such a trial should include a hydration protocol and routine use of low or iso-osmolar nonionic contrast media, since these are the current standard of care. Finally, although absolute or relative changes in serum creatinine level are reasonable surrogate end points, more important and clinically relevant end points should also be primarily addressed.6,48,49
Correspondence: William A. Ghali, MD, MPH, Faculty of Medicine, University of Calgary, 3330 Hospital Dr NW, Calgary, Alberta, Canada T2N 4N1 (firstname.lastname@example.org).
Accepted for Publication: December 24, 2004.
Financial Disclosure: None.
Funding/Support: Dr Bagshaw is supported by a Canadian Institutes for Health Research Canada Graduate Scholarship Masters Award. Dr Ghali is supported by a Government of Canada Research Chair in Health Services Research and by a Health Scholar Award from the Alberta Heritage Foundation for Medical Research.