Abstract
OBJECTIVE.
The purpose of this study was to assess the mechanism by which aspirin therapy improves survival when combined with transarterial chemoembolization or transarterial embolization (TAE) for hepatocellular carcinoma (HCC).
MATERIALS AND METHODS.
A retrospective review included 304 patients with HCC who were treated with TAE. The patients were divided into two groups on the basis of whether the patient took aspirin (n = 42) or did not take aspirin (n = 262) at the time of initial TAE. For each patient, response of embolized tumors, time to progression, initial site of progression, survival time, and liver function test results before and after embolization were evaluated.
RESULTS.
Patients taking aspirin and those not taking aspirin at the time of initial TAE for HCC had no difference in initial response rate (88% vs 90% complete response or partial response, p = 0.59), median time to progression (6.2 vs 5.2 months, p = 0.42), initial site of progression (p = 0.77), or fraction of patients dying with disease progression (88% vs 89%, p = 1.00). Before embolization, there was no difference in mean bilirubin level (0.8 vs 0.9 mg/dL, p = 0.11) for patients taking versus not taking aspirin. Among patients taking aspirin, bilirubin level was significantly lower 1 day (0.9 vs 1.3, p < 0.001), 1 month (0.9 vs 1.2, p = 0.048), and 1 year (0.8 vs 1.0, p = 0.021) after embolization. The median overall survival period after initial embolization was longer for patients taking aspirin (57 vs 23 months, p = 0.008).
CONCLUSION.
Aspirin use is associated with improved liver function test results and survival after TAE for HCC. It is not associated with differences in response or time to progression.
Keywords: aspirin, embolization, hepatocellular carcinoma, hepatoprotective, liver function
Retrospective studies have shown that aspirin use is associated with improved survival when combined with transarterial chemoembolization [1] or transarterial embolization (TAE) [2] for hepatocellular carcinoma (HCC). Several theories have been proposed to explain these results [2]. First, aspirin is antiinflammatory. Chronic inflammation in the liver can result in cirrhosis and development of HCC. Retrospective studies have shown that aspirin reduces the rate of death of chronic liver disease [3] and reduces the risk of development of HCC [3]. Second, aspirin is antiangiogenic. It inhibits hypoxia-induced angiogenesis in in vitro models [4, 5], which might prevent TAE-induced ischemia from promoting angiogenesis and growth of the residual viable tumor [6, 7]. Third, aspirin is antiglycolytic. It decreases glucose consumption and the viability of tumor cells in vitro [8], which might make tumor cells less likely to survive TAE-induced ischemia [9].
To distinguish between these three mechanisms, we can examine response, recurrence, and liver function after embolization of HCC. If the antiinflammatory mechanism is dominant, we would expect patients taking aspirin to have better liver function and to be less likely to have new HCC develop. If the antiangiogenic mechanism is dominant, we would expect patients taking aspirin to have a longer time to progression after embolization. If the antiglycolytic mechanism is dominant, we would expect patients taking aspirin to have a higher rate of response to embolization.
The purpose of this study was to retrospectively compare response, time to progression, and liver function after embolization of HCC in patients taking versus those not taking aspirin to gain insights into the mechanism of improved survival among patients taking aspirin.
Materials and Methods
Patient Selection
This HIPAA-compliant retrospective study received institutional review board approval. A previously reported retrospective review [2] had identified 304 consecutively registered patients with HCC (232 [76%] men, 72 [24%] women; mean age at embolization, 69 years) who were treated with TAE between June 2009 and April 2016. Further analysis of this group of patients was conducted in the current study. Patients with prior ablation or radioembolization were excluded. Particle embolization was performed to stasis according to a standard published protocol [10].
Aspirin Use
According to hospital policy, the outpatient medication list (including over-the-counter medications) was reviewed with the patient at each outpatient or inpatient visit. Patients who reported taking aspirin 7 days before through 30 days after initial TAE (n = 42) were compared with patients who did not take aspirin (n = 262) during the same time interval. Aspirin doses were 81 mg orally daily (35 patients), 325 mg orally daily (five patients), 162 mg orally daily (one patient), and 81 mg orally every other day (one patient).
Retrospective Chart Review
For each patient, the following information was collected: response of embolized tumors, time to progression after initial embolization, initial site of progression (treated tumor, untreated liver tumor, or extrahepatic tumor), survival time, cause of death, and liver function test results before and after embolization. Response of embolized tumors was evaluated by means of the modified Response Evaluation Criteria in Solid Tumors (mRECIST) [11] at contrast-enhanced CT or MRI performed 1 month after initial embolization. Time to progression at any site was evaluated with mRECIST after completion of the originally planned staged treatment (for example, right and left lobe embolization). The initial site of progression was classified as follows: progression of a treated liver tumor, new or growing tumor in the liver without progression of any treated tumor, or extrahepatic progression without intrahepatic progression. Images were reviewed by a board-certified radiologist with 8 years of experience in abdominal imaging who was blinded to the aspirin status of each patient at image review.
Mean liver function values (total bilirubin, aspartate aminotransferase [AST], alanine aminotransferase [ALT], albumin, and international normalized ratio [INR]) at each time point after embolization were determined by examining the most recent laboratory values of patients who were still undergoing follow-up laboratory tests. Preprocedure liver function tests were performed within 30 days before the procedure. In the calculation of mean INR, 29 patients who underwent outpatient anticoagulation were excluded. Patients who received prophylactic inpatient heparin therapy were not excluded.
Statistical Analysis
Kaplan-Meier curves were compared by logrank test. Proportions were compared by Fisher exact test or chi-square test. Laboratory values were compared by two-tailed t tests and repeated-measures ANOVA. Statistical tests were performed with Mathematica 9 (Wolfram) and Excel 2016 (Microsoft). Values of p < 0.05 were considered significant.
Results
Survival
Prior analysis of these patients [2] showed that the median overall survival time after initial embolization was longer for patients taking aspirin (57 vs 23 months, p = 0.008). Before embolization, the group taking aspirin and the group not taking aspirin had no difference in American Joint Committee on Cancer (AJCC) stage, Child-Pugh score, cause of underlying liver disease, Eastern Cooperative Oncology Group performance status, prior sorafenib therapy, or prior liver resection [2]. There also was no difference in risk of postoperative death due to comorbidities (Charlson comorbidity index) for patients taking versus those not taking aspirin [2]. Indications for taking aspirin were coronary artery disease (15 patients), cardioprotective effect in the absence of documented coronary artery disease (one patient), atrial fibrillation (one patient), peripheral arterial disease (six patients), cerebrovascular disease (two patients), multiple vascular indications (five patients), arthritis (four patients), and unknown (eight patients).
Response
Four patients had no follow-up images, and two patients had no measurable lesions according to mRECIST. The other 298 patients had no difference in mRECIST response between the aspirin and no aspirin groups (88% vs 90% complete or partial response, p = 0.59 by Fisher exact test (Table 1).
TABLE 1:
Response to Embolization of Hepatocellular Carcinoma 1 Month After Embolization
| mRECIST Category | Aspirin | No Aspirin |
|---|---|---|
| Complete response | 20 (48) | 111 (43) |
| Partial response | 17 (40) | 121 (47) |
| Stable disease | 0 (0) | 11 (4) |
| Progressive disease | 5 (12) | 13 (5) |
| Total | 42 | 256 |
Note—There was no significant difference between the aspirin group and the no aspirin group (p = 0.59). Values in parentheses are percentages, which owing to rounding do not total 100 in the no aspirin group. mRECIST = modified Response Evaluation Criteria in Solid Tumors.
Progression
Median times to progression were 5.2 months in the no aspirin group and 6.2 months in the aspirin group (p = 0.42 by logrank test) (Fig. 1). Among patients with progressive disease, there was no difference in the initial site of progression (p = 0.77 by chi-square test) (Table 2).
Fig. 1—
Graph shows time to progression (any site) after initial transarterial embolization for hepatocellular carcinoma (p = 0.42).
TABLE 2:
Initial Site of Progression After Embolization of Hepatocellular Carcinoma
| Site | Aspirin | No Aspirin |
|---|---|---|
| Treated lesion | 21 (52) | 102 (48) |
| Liver | 16 (40) | 88 (41) |
| Extrahepatic | 3 (8) | 23 (11) |
| Total | 40 | 213 |
Note—No significant difference was found between the aspirin group and the no aspirin group (p = 0.77). Values in parentheses are percentages.
Cause of Death
To determine the fraction of patients who died with disease progression, we examined patients who died and had either progression at imaging or no progression at imaging within 6 months of death. In the no aspirin group, 127 of 142 (89%) patients died with evidence of HCC progression at imaging. In the aspirin group, seven of eight (88%) patients died with HCC progression at imaging (p = 1.00 by Fisher exact test). For the 15 (11%) patients in the no aspirin group who died without documented HCC progression, the causes of death were liver failure (three patients), pneumonia (two patients), bleeding varices (one patient), and unknown (nine patients). For the one patient in the aspirin group who died without documented HCC progression, the cause of death was liver failure. None of the patients died of myocardial infarction or stroke.
Liver Function
Before embolization, there was no difference in mean bilirubin level (0.8 vs 0.9 mg/dL, p = 0.11) for patients taking versus not taking aspirin (Table 3 and Figs. 2 and 3). The bilirubin level was significantly lower for patients taking aspirin when it was measured 1 day (0.9 vs 1.3, p < 0.001), 1 month (0.9 vs 1.2, p = 0.048), and 1 year (0.8 vs 1.0, p = 0.021) after embolization.
TABLE 3:
Laboratory Values Before and After Embolization of Hepatocellular Carcinoma Among Patients Taking and Not Taking Aspirin
| No Aspirin |
Aspirin |
|||||
|---|---|---|---|---|---|---|
| Time After Initial Embolization (d) | Mean | SD | No. | Mean | SD | No. |
| Bilirubin (mg/dL) | ||||||
| Before | 0.94 | 0.57 | 262 | 0.77 | 0.65 | 42 |
| 1 | 1.22 | 0.88 | 262 | 0.87 | 0.39 | 42 |
| 30 | 1.19 | 0.96 | 255 | 0.94 | 0.69 | 42 |
| 120 | 1.1 | 1.01 | 233 | 0.83 | 0.56 | 41 |
| 365 | 1.02 | 0.76 | 172 | 0.75 | 0.56 | 35 |
| Albumin (g/dL) | ||||||
| Before | 3.9 | 0.47 | 262 | 4.03 | 0.39 | 42 |
| 1 | 3.66 | 0.45 | 262 | 3.76 | 0.36 | 42 |
| 30 | 3.54 | 0.52 | 255 | 3.71 | 0.34 | 42 |
| 120 | 3.7 | 0.50 | 233 | 3.93 | 0.44 | 41 |
| 365 | 3.76 | 0.52 | 172 | 3.91 | 0.45 | 35 |
| International normalized ratioa | ||||||
| Before | 1.13 | 0.13 | 242 | 1.09 | 0.12 | 33 |
| 1 | 1.17 | 0.15 | 242 | 1.14 | 0.15 | 33 |
| 30 | 1.21 | 0.18 | 235 | 1.15 | 0.16 | 33 |
| 120 | 1.17 | 0.19 | 214 | 1.12 | 0.14 | 32 |
| 365 | 1.13 | 0.14 | 159 | 1.13 | 0.13 | 27 |
| Aspartate transaminase (U/L) | ||||||
| Before | 72 | 57 | 262 | 53 | 32 | 42 |
| 1 | 386 | 610 | 262 | 287 | 562 | 42 |
| 30 | 115 | 161 | 255 | 75 | 73 | 42 |
| 120 | 94 | 147 | 233 | 72 | 78 | 41 |
| 365 | 76 | 111 | 172 | 64 | 46 | 35 |
| Alanine transaminase (U/L) | ||||||
| Before | 62 | 49 | 262 | 50 | 31 | 42 |
| 1 | 254 | 423 | 262 | 192 | 336 | 42 |
| 30 | 112 | 161 | 255 | 82 | 77 | 42 |
| 120 | 79 | 119 | 233 | 66 | 68 | 41 |
| 365 | 61 | 71 | 172 | 59 | 41 | 35 |
Note—All patients had laboratory values before and 1 day after embolization.
Includes only patients not undergoing anticoagulation.
Fig. 2—
Graphs show mean laboratory values before (Pre) and after initial embolization of hepatocellular carcinoma. Results of t tests comparing groups taking aspirin and groups not taking aspirin are as follows: not significant (NS; p > 0.05), * (p ≤ 0.05), ** (p ≤ 0.01), *** (p ≤ 0.001).
A, Total bilirubin level.
B, Albumin level.
C, Aspartate transaminase (AST) level.
D, Alanine transaminase (ALT) level.
E, International normalized ratio (INR).
Fig. 3—
Tukey box plots show laboratory values before (Pre) and after initial embolization of hepatocellular carcinoma (median, quartiles, and outliers). Whiskers show lowest value within 1.5 interquartile range (IQR) of lower quartile and highest value within 1.5 IQR of upper quartile. Results of statistical tests are shown in Figure 2 and Table 4.
A, Total bilirubin level.
B, Albumin level.
C, Aspartate transaminase (AST) level.
D, Alanine transaminase (ALT) level.
E, International normalized ratio (INR).
Before embolization, there was no difference in albumin level or INR between the group taking aspirin and the group not taking aspirin. One and 4 months after embolization, the albumin level was significantly worse in the group not taking aspirin. The AST level was significantly worse in the group not taking aspirin both before and 1 month after embolization. The ALT level before embolization was significantly worse in the group not taking aspirin. Repeated-measures ANOVA confirmed that the changes in liver function test results after embolization were different for the groups taking aspirin and not taking aspirin (Table 4).
TABLE 4:
Repeated-Measures ANOVA Results
| p | |||||
|---|---|---|---|---|---|
| Variable | Bilirubin | Albumin | Aspartate Transaminase | Alanine Transaminase | INR |
| Aspirin | ≤ 0.05 | NS | NS | NS | NS |
| Time | ≤ 0.05 | ≤ 0.001 | ≤ 0.001 | ≤ 0.001 | ≤ 0.001 |
| Aspirin × time | ≤ 0.05 | ≤ 0.001 | ≤ 0.001 | ≤ 0.001 | ≤ 0.001 |
Note—Results show that liver function test results change over time after embolization (Time), and that the liver function test result changes over time are significantly different in the aspirin group compared with the no aspirin group (Aspirin × time). NS = not significant (p>0.05).
Discussion
Aspirin therapy is associated with both improved results of liver function tests and improved survival after TAE for HCC. It is not associated with any differences in response, time to progression, or initial site of progression (arguing against a direct antitumor effect). None of the patients died of myocardial infarction or stroke (arguing against a cardioprotective effect causing the improved survival). There was no difference in the predicted risk of postoperative mortality due to age or comorbidities in the group taking versus the group not taking aspirin according to the Charlson comorbidity index (arguing against preferential use of aspirin by healthier patients). The groups taking aspirin and not taking aspirin also had no difference in AJCC stage, Child-Pugh score, or cause of underlying liver disease before TAE. These results suggest that aspirin is hepatoprotective for patients with HCC treated with embolization.
Previously published data also support our theory that aspirin improves survival by reducing liver inflammation and thus improving liver function [2, 3, 12, 13]. Other nonsteroidal antiinflammatory drugs (NSAIDs), most commonly, ibuprofen and naproxen, were also associated with improved survival when taken at the time of embolization for HCC [2]. Aspirin was not associated with survival differences after locoregional therapy for neuroendocrine tumors or colorectal liver metastases [2], probably because patients with these cancers are less likely to have cirrhosis. Retrospective studies have shown that patients with chronic liver disease who take aspirin have decreased liver fibrosis [12] and decreased risk of death of chronic liver disease [3]. An animal study [13] showed that aspirin and clopidogrel therapy decreased liver inflammation, liver fibrosis, and development of HCC in a mouse model of chronic hepatitis B.
Although the differences in liver function test results in the groups taking and not taking aspirin were small, standard biochemical liver function tests are insensitive to early cirrhotic changes. Perfusion imaging can be used to detect early cirrhotic changes that are predictive of survival but that are not reflected in the Child-Pugh score [14]. Results of biochemical liver function tests are not accurate predictors of the finding of fibrosis at liver biopsy [15]. Thus, a normal bilirubin level does not mean that the liver is normal; small changes in biochemical liver function test results might underestimate the degree of liver injury after embolization.
Improved liver function in the aspirin group was persistent for at least 1 year after embolization. A previous analysis by our group [16] showed that the patients taking aspirin were more likely to remain candidates for additional locoregional therapy. In the current study, the aspirin group had better AST results both before and after embolization, but bilirubin and albumin results were better only after embolization. This finding suggests that aspirin can improve liver function in patients with chronic liver disease but that the effect is more pronounced after TAE.
The lack of difference in response and time to progression argues against a substantial amount of antiglycolytic or antiangiogenic activity of aspirin in this patient population, most of whom took 81 mg of aspirin daily. In vitro experiments have shown antiangiogenic [4, 5] and antiglycolytic [8] effects of aspirin at concentrations higher than the typical peak plasma concentration of aspirin after an oral dose of 320 mg [17]. In addition, a case report [18] described HCC that responded to celecoxib, which is a selective cyclooxygenase 2 (COX-2) inhibitor (aspirin inhibits COX-1 and COX-2). Future experiments should examine the optimal dose, route (oral or intraarterial), and type of NSAID to achieve the optimal combination of antiinflammatory, antiglycolytic, and antiangiogenic effects.
The main limitation of this study was that it was retrospective; thus, a confounding variable may account for the improved survival among patients taking aspirin. However, the groups taking aspirin and not taking aspirin had comparable liver function, AJCC stage, comorbidities, and other clinical characteristics before embolization. In addition, published results of animal studies and retrospective human data support the connection between NSAID use and decreased liver inflammation. Aspirin and β-blockers are typically taken together [2], raising the possibility that β-blockers, not aspirin, could be responsible for the improved survival. However, β-blockers reduce blood flow in the portal vein and thus would not be expected to improve liver function [19]. The second limitation of this study was that the association between aspirin use and survival was identified in a prior study in which 29 different medications or medication classes were examined [2]. Although the connection between aspirin and survival remained significant after Bonferroni correction for multiple comparisons, the magnitude of the survival benefit was probably overestimated [20]. The third limitation of this study was that there were too few patients to assess the optimal dose of aspirin and the dosage timing or to analyze the effect of aspirin on subgroups of patients (such as those with hepatitis B versus hepatitis C virus infection).
Conclusion
Aspirin use is associated with improved liver function test results and survival after TAE for HCC. It is not associated with differences in response or time to progression. The results should be confirmed in a randomized controlled trial.
Acknowledgment
We thank Daniel Kelly for writing queries for extracting information from the clinical database at our hospital.
Supported in part by an NIH/NCI Cancer Center support grant (P30 CA008748).
F. E. Boas is a cofounder of Claripacs, LLC; has received research funding and support for research meetings from Guerbet; has received research support from GE Healthcare; has received research supplies from Bayer HealthCare; has received a research grant and speaker fees from the Society of Interventional Oncology, sponsored by Guerbet; and is an investor in Labdoor, Qventus, CloudMedx, and Notable Laboratories. C. T. Sofocleous is a consultant for Sirtex and Neuwave/Johnson & Johnson and receives research support from TheraSphere/Biocompatibles. S. B. Solomon is a consultant to BTG, Johnson & Johnson, Adgero, Aperture Medical, XACT, and Innoblative; has received research grants from GE Healthcare, AngioDynamics, and Elesta; and is a shareholder in Johnson & Johnson, Aperture Medical, Aspire Bariatrics, Immunomedics, Progenics, Motus GI, and Impulse Dynamics.
Footnotes
Based on a presentation at the Society of Interventional Radiology 2018 annual meeting, Los Angeles, CA.
References
- 1.Li JH, Wang Y, Xie XY, et al. Aspirin in combination with TACE in treatment of unresectable HCC: a matched-pairs analysis. Am J Cancer Res 2016; 6:2109–2116 [PMC free article] [PubMed] [Google Scholar]
- 2.Boas FE, Ziv E, Yarmohammadi H, et al. Adjuvant medications that improve survival after locoregional therapy. J Vasc Interv Radiol 2017; 28:971–977 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sahasrabuddhe VV, Gunja MZ, Graubard BI, et al. Nonsteroidal anti-inflammatory drug use, chronic liver disease, and hepatocellular carcinoma. J Natl Cancer Inst 2012; 104:1808–1814 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Khaidakov M, Szwedo J, Mitra S, et al. Antiangiogenic and antimitotic effects of aspirin in hypoxia–reoxygenation modulation of the LOX-1-NADPH oxidase axis as a potential mechanism. J Cardiovasc Pharmacol 2010; 56:635–641 [DOI] [PubMed] [Google Scholar]
- 5.Borthwick GM, Johnson AS, Partington M, Burn J, Wilson R, Arthur HM. Therapeutic levels of aspirin and salicylate directly inhibit a model of angiogenesis through a Cox-independent mechanism. FASEB J 2006; 20:2009–2016 [DOI] [PubMed] [Google Scholar]
- 6.Liang B, Zheng CS, Feng GS, et al. Correlation of hypoxia-inducible factor 1alpha with angiogenesis in liver tumors after transcatheter arterial embolization in an animal model. Cardiovasc Intervent Radiol 2010; 33:806–812 [DOI] [PubMed] [Google Scholar]
- 7.Wang B, Xu H, Gao ZQ, Ning HF, Sun YQ, Cao GW. Increased expression of vascular endothelial growth factor in hepatocellular carcinoma after transcatheter arterial chemoembolization. Acta Radiol 2008; 49:523–529 [DOI] [PubMed] [Google Scholar]
- 8.Spitz GA, Furtado CM, Sola-Penna M, Zancan P. Acetylsalicylic acid and salicylic acid decrease tumor cell viability and glucose metabolism modulating 6-phosphofructo-1-kinase structure and activity. Biochem Pharmacol 2009; 77:46–53 [DOI] [PubMed] [Google Scholar]
- 9.Yarmohammadi H, Wilkins LR, Erinjeri JP, et al. Efficiency of combined blocking of aerobic and glycolytic metabolism pathways in treatment of N1-S1 hepatocellular carcinoma in a rat model. J Cancer Res Ther 2017; 13:533–537 [DOI] [PubMed] [Google Scholar]
- 10.Brown KT. Bland embolization for hepatic malignancies. In: Mauro MA, Murphy KP, Thomson KR, Venbrux AC, Morgan RA, eds. Image-guided interventions Philadelphia, PA: Saunders Elsevier, 2014:467–475 [Google Scholar]
- 11.Lencioni R, Llovet JM. Modified RECIST (mRECIST) assessment for hepatocellular carcinoma. Semin Liver Dis 2010; 30:52–60 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Jiang ZG, Feldbrügge L, Tapper EB, et al. Aspirin use is associated with lower indices of liver fibrosis among adults in the United States. Aliment Pharmacol Ther 2016; 43:734–743 [DOI] [PubMed] [Google Scholar]
- 13.Sitia G, Aiolfi R, Di Lucia P, et al. Antiplatelet therapy prevents hepatocellular carcinoma and improves survival in a mouse model of chronic hepatitis B. Proc Natl Acad Sci USA 2012; 109:E2165–E2172 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Borgheresi A, Gonzalez-Aguirre A, Brown KT, et al. Does enhancement or perfusion on preprocedure CT predict outcomes after embolization of hepatocellular carcinoma? Acad Radiol 2018; 25:1588–1594 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rossi E, Adams L, Prins A, et al. Validation of the FibroTest biochemical markers score in assessing liver fibrosis in hepatitis C patients. Clin Chem 2003; 49:450–454 [DOI] [PubMed] [Google Scholar]
- 16.Boas FE, Ziv E, Yarmohammadi H, et al. Reply to “Adjuvant medications that improve survival after locoregional therapy.” J Vasc Interv Radiol 2017; 28:1335–1336 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cerletti C, Bonati M, del Maschio A, et al. Plasma levels of salicylate and aspirin in healthy volunteers: relevance to drug interaction on platelet function. J Lab Clin Med 1984; 103:869–877 [PubMed] [Google Scholar]
- 18.Malka D, Pacault V, De Baere T, Ducreux M, Boige V. Antitumoral effect of celecoxib in hepatocellular carcinoma. J Clin Oncol 2005; 23:4805–4806 [DOI] [PubMed] [Google Scholar]
- 19.Garcia-Tsao G, Abraldes JG, Berzigotti A, Bosch J. Portal hypertensive bleeding in cirrhosis: risk stratification, diagnosis, and management: 2016 practice guidance by the American Association for the Study of Liver Diseases. Hepatology 2017; 65:310–335 [DOI] [PubMed] [Google Scholar]
- 20.Forstmeier W, Schielzeth H. Cryptic multiple hypotheses testing in linear models: overestimated effect sizes and the winner’s curse. Behav Ecol Sociobiol 2011; 65:47–55 [DOI] [PMC free article] [PubMed] [Google Scholar]