In re Lipitor (Atorvastatin Calcium) Marketing, Sales Practices & Products Liability Litigation

Citations: 185 F. Supp. 3d 761; 2016 U.S. Dist. LEXIS 68829; 2016 WL 2840215Docket: MDL No. 2:14-mn-02502-RMG

Court: District Court, D. South Carolina; May 6, 2016; Federal District Court

SummaryOriginalCytator Case TreatmentBackward CitatorAnalysis

EnglishEspañolSimplified EnglishEspañol Fácil

CASE MANAGEMENT ORDER NO. 72 addresses Pfizer's Motion to Exclude Expert Testimony regarding Lipitor's effectiveness for primary prevention in women. The motion is partially granted and partially denied. All plaintiffs in the MDL are women who allege that Lipitor, used for lowering LDL cholesterol and triglycerides or for primary prevention of cardiovascular disease (CVD), caused Type 2 diabetes and that Pfizer failed to provide adequate warnings about this risk. Plaintiffs argue Lipitor is not effective for women in preventing CVD and that it was negligently designed, with insufficient warnings about its risks and ineffectiveness.

The Court will evaluate expert opinions asserting that Lipitor lacks evidence of effectiveness for primary prevention in women, contrasting with the plaintiffs' Master Complaint, which explicitly claims it is ineffective. Although the plaintiffs assert their experts do not claim Lipitor should never have been approved for women, one expert indicated that it ought to be considered for off-label use only. Pfizer contends all opinions should be excluded under Federal Rule of Evidence 702 and claims that the efficacy allegations are preempted by federal law concerning product labeling and promotion. The Court will first address the preemption issue before considering Rule 702 and Daubert standards.

Congress has the authority to preempt state law under the Constitution, as established in Crosby v. Nat’l Foreign Trade Council. Impossibility preemption occurs when compliance with both state and federal law is unachievable. The Defendant argues that it could not meet state law obligations regarding label changes while adhering to federal labeling requirements. The Court must assess whether the Defendant could modify its label to reflect a lack of benefit for women, as alleged by Plaintiffs, without conflicting with federal law.

Under the Federal Drug and Cosmetic Act (FDCA), drug manufacturers must obtain FDA approval before marketing a drug. This involves submitting a new-drug application (NDA) or supplemental new-drug application (sNDA), including comprehensive safety and effectiveness studies. The FDA reviews proposed labeling and must conclude that the drug’s labeling is not misleading and that substantial evidence supports its effectiveness. The FDA has the discretion to determine that one well-controlled clinical investigation, supplemented by additional evidence, can suffice for establishing drug effectiveness. Upon FDA approval, manufacturers can market the drug using the approved label without violating federal law.

Specifically, Lipitor was approved by the FDA in 1996 for cholesterol reduction, with subsequent approvals for additional uses. The 2004 indication for Lipitor is relevant here, as it outlines its use in patients without evident coronary heart disease but with risk factors for cardiovascular disease, aimed at reducing risks of myocardial infarction, stroke, revascularization procedures, and angina.

The FDA approved Lipitor based on the ASCOT clinical trial, with a general approval not specifying gender. However, the ASCOT study's label indicated inconclusive results for women due to a limited number of events. A brand-name drug manufacturer can modify a drug's label either by obtaining FDA approval or through the Changes Being Effected (CBE) regulations. According to the U.S. Supreme Court, if a manufacturer must secure FDA approval to comply with state law, state law claims are preempted. The key issue is whether the Defendant could have corrected the labeling deficiency using the CBE regulation without prior FDA approval. Under the CBE regulation, manufacturers can independently change a label to reflect new information, such as adding or strengthening warnings or deleting misleading claims. Plaintiffs argue that the Defendant should have altered Lipitor’s label regarding efficacy under subsection (D) of this regulation. Newly acquired information, defined as data not previously submitted to the FDA, can justify label changes. Claims for changes based on previously submitted information are preempted, while claims based on new information are not. This distinction allows states to act on new information not considered by the FDA, while the FDA retains authority over safety and efficacy based on information available at the time of marketing.

Preemption is linked to the availability of procedures for drug label changes, allowing the FDA to manage new drug launches while permitting states to require manufacturers to address information not reviewed by the FDA. In the case of Lipitor, the FDA approved its label based on the ASCOT trial, which indicated inconclusive results for women due to a small event count. Claims from Plaintiffs regarding the need for different labeling based on ASCOT data are preempted, as they did not identify any new analyses that would warrant a label change. The only analysis mentioned was conducted by Dr. Wells for this litigation, and while the CASHMERE study was presented as new information, it is not preempted since it pertains to newly acquired data. However, the relevance of CASHMERE to primary prevention indications is debated, but this issue pertains to evidentiary sufficiency rather than preemption. Plaintiffs assert that accumulating information about diabetes risks from statins should prompt changes in the Lipitor label regarding efficacy. The Court disagrees, stating that changes concerning efficacy must be based on newly acquired information specifically relating to efficacy, not general safety data. The regulatory framework stipulates that label modifications can only occur when new information proves prior claims of effectiveness to be false or unsupported. Thus, the CBE regulation allows for changes related to efficacy only when directly supported by new efficacy-related information.

Claims by Plaintiffs regarding a state law duty for Defendant to modify Lipitor's label to include different statements about its efficacy for primary prevention in women, based on ASCOT data or diabetes risk information, are preempted. However, claims based on CASHMERE or other newly acquired information are not preempted. Additionally, Plaintiffs assert that Defendant should have altered its advertising and promotion of Lipitor concerning efficacy, but Defendant argues these claims are also preempted. Any claim suggesting that Defendant should stop selling Lipitor to women for primary prevention is preempted, as established by U.S. Supreme Court precedent (Mut. Pharm. Co. v. Bartlett). A state law judgment claiming Pfizer's advertising contained falsehoods not reflected on the Lipitor label does not necessitate stopping sales. The court examines whether state law claims related to Defendant’s promotion and advertising are preempted by the Federal Food, Drug, and Cosmetic Act (FDCA). The FDCA broadly defines labeling, which includes all written or graphic materials accompanying a drug, such as advertising and promotional materials. This definition encompasses a wide range of communications related to the drug, including brochures and 'Dear Doctor' letters, all of which are considered labeling under FDCA regulations. Thus, virtually all communications with healthcare professionals about a drug fall under this labeling definition.

Plaintiffs' claims regarding promotional materials directed at medical professionals are preempted to the same extent as their claims based on the Lipitor label, as these materials are classified as part of the labeling. However, not all advertising materials are considered labeling. According to 21 U.S.C. 352(n), while promotional materials for medical professionals fall under labeling, advertisements aimed at the general public do not and can be modified without adhering to the Changes Being Effected (CBE) regulation. The defendant has not provided evidence that a brand-name drug manufacturer is prohibited from altering public advertisements independent of label changes. Although such advertisements are heavily regulated, the defendant has not demonstrated any conflict with state law. The court notes that federal regulations mandate that all drug advertisements must accurately convey effectiveness, but the defendant has not claimed that this requirement conflicts with state law obligations.

Regarding the admissibility of scientific testimony under Federal Rule of Evidence 702, the trial judge must ensure that the evidence is relevant and reliable. The court must assess whether the testimony is derived from reliable methods, whether these methods have been properly applied to the case facts, and whether there is sufficient supporting data. Key factors for this evaluation include the testability of the theory, peer review status, error rates, standard maintenance, and general acceptance within the scientific community.

Factors influencing the admissibility of expert testimony are not exhaustive and include considerations such as whether the expert formulated their opinions specifically for litigation or through independent research. Courts assess if experts have addressed conflicting literature and require sufficient factual or data support for their opinions, as mandated by Rule 702. If there is a significant analytical gap between the data and the expert's opinion, the court may exclude that testimony. The proponent of expert testimony bears the burden of establishing its admissibility by a preponderance of the evidence. The Daubert inquiry balances the liberalization of expert evidence introduction with the need for rigorous reliability analysis due to the potential for expert testimony to mislead.

Dr. Wells, a statistician, was tasked with reviewing the statistical aspects of atorvastatin studies related to gender heterogeneity and efficacy. The ASCOT-LLA study evaluated the efficacy of Lipitor in preventing coronary heart disease (CHD) in patients with high blood pressure but normal blood lipids. The study involved 10,305 participants, aged 40-79, with multiple cardiovascular risk factors. It was terminated early due to significant reductions in CHD events and stroke incidence among participants taking atorvastatin compared to those on placebo. Notably, of the study participants, only 1,942 were women, and while the overall findings were positive, no significant benefits were observed in the female subgroup, with similar primary endpoint events occurring in both women on atorvastatin and those on placebo.

The authors noted a limited benefit of atorvastatin on the primary endpoint for women, possibly due to the low number of events (36 occurrences). In terms of secondary endpoints, while women taking Lipitor showed a lower rate of total cardiovascular and coronary events, the differences were not statistically significant. Gender heterogeneity tests revealed no significant interaction between sex and treatment effects on the primary endpoint, indicating that the outcomes were similar for both genders. Contrarily, Dr. Wells conducted a separate analysis suggesting significant gender heterogeneity across all endpoints, claiming that treatment effectiveness varied by gender. He reached this conclusion using the Aalen’s linear hazards model instead of the prespecified Cox proportional hazards model, which he argued was inappropriate due to its assumption of a constant hazard ratio over time. Dr. Wells’ tests indicated that this assumption did not hold for the ASCOT data, leading him to assert that the Aalen model revealed significant differences in outcomes between men and women. Pfizer contested Dr. Wells' findings, arguing that the Cox model is the standard for such analyses and that Wells’ reanalysis is litigation-driven. However, the Court deemed Dr. Wells’ reanalysis admissible under Rule 702, noting the need for valid justification in reanalyses of published data. Dr. Wells successfully provided such justification by demonstrating the inapplicability of the Cox model to the ASCOT data.

Dr. Wells found that the Cox model's assumptions were violated and opted for the Aalen model, which is recognized in peer-reviewed literature and not a novel creation for this case. He noted that he was unique in assessing the proportional hazards assumption, providing a valid basis for his reanalysis. The court rejected the defendant's motion to exclude this reanalysis under Rule 702, acknowledging Wells' justification for using the Aalen model. 

In evaluating primary prevention efficacy studies, plaintiffs' experts claim there is insufficient evidence supporting statins' effectiveness for women. The 2008 JUPITER study involved 17,802 participants, including 6,801 women (38.2%), and demonstrated significant cardiovascular benefits from rosuvastatin, with similar relative hazard reductions for both genders (46% for women; 42% for men). A separate analysis reported a significant hazard ratio of .54 for women concerning primary endpoints. 

The 2004 CARDS study examined atorvastatin's efficacy in 2,838 diabetic patients with normal LDL levels, including 32% women. The study, terminated early due to positive results, found a 37% reduction in major cardiovascular events associated with atorvastatin, with no significant impact from sex on treatment effects.

Mora (2010) performed meta-analyses on statin therapy for the primary prevention of cardiovascular disease (CVD) in women, analyzing trials such as AFGAPS/TexCAPS, MEGA, and JUPITER. Results indicated that statin therapy reduced CVD risk by approximately one-third compared to placebo, with a relative risk ratio of 0.63 (CI: 0.49-0.82, p<0.001) and no gender heterogeneity. When including two additional trials, ASCOT and ALLHAT-LLT, the significance diminished (relative risk 0.79, CI: 0.59-1.05, p=0.1), but gender heterogeneity remained absent. Including two more trials, HPS and PROSPER, restored statistical significance (relative risk 0.82, CI: 0.69-0.98, p=0.03), suggesting prior insignificance was due to low event numbers among women.

Kostis (2012) aimed to evaluate statin effects on cardiovascular events between sexes, analyzing data from 18 randomized trials. Statistically significant benefits of statins were observed for both sexes, with women showing an Odds Ratio of 0.81 (CI: 0.75-0.89, p<0.0001) and men 0.77 (CI: 0.71-0.83, p<0.0001). In primary prevention, the Odds Ratio for women was 0.85 (CI: 0.75-0.98), and for secondary prevention, it was 0.78 (CI: 0.70-0.88). All-cause mortality was lower in women on statins, with an Odds Ratio of 0.90 across all studies and 0.87 for primary prevention studies. Analyses by risk level demonstrated significant benefits for all risk categories, with lower-risk women experiencing greater benefits.

The CTT Collaboration (2015) conducted a meta-analysis comparing statin effects between genders across 27 trials, focusing on cardiovascular risk rather than primary/secondary prevention classification.

The authors conducted a meta-analysis comparing statin efficacy by sex, using 99% confidence intervals. They assert this is the largest such analysis and the only one to adjust for detailed cardiovascular risk factors. The study reveals that women have a lower baseline cardiovascular disease risk, but in individuals of equivalent risk, statin therapy is equally effective for preventing major vascular events in both sexes. The analysis of 27 trials showed that a 1.0 mmol/L reduction in LDL cholesterol corresponds to a 21% reduction in major vascular events for both men and women. While there was a slight difference in the proportional reductions for women compared to men in 22 trials comparing statins to a placebo, these results remained statistically significant for both genders. After adjusting for non-sex differences in baseline risk factors, any observed differences between sexes largely disappeared. The authors concluded that statin therapy yields similar proportional and absolute effects on major vascular events and mortality among men and women at comparable risk levels, particularly in low-risk populations.

Dr. Wells critiques the evidence supporting the notion that statins provide primary cardioprotection for women, claiming no proper meta-analysis shows statistically significant evidence for this claim. He argues that the misperception of statin efficacy in women stems from non-reporting of randomized controlled trials (RCTs) and improper trial pooling. Dr. Wells maintains that there is no evidence supporting the use of statins for primary prevention in women, positing that even one contradictory study would undermine his assertion. However, it is noted that the JUPITER study does provide statistically significant evidence of primary prevention benefits for women, a fact Dr. Wells acknowledges but does not dispute as a primary prevention study. He suggests that other meta-analyses would not yield significant findings for women if JUPITER was excluded.

Dr. Wells acknowledges the findings of the CARDS study, which shows cardiovascular benefits from statins for women, particularly in the context of primary prevention. He classifies diabetes as a "risk equivalent" to coronary heart disease (CHD) and suggests that diabetic women benefit from statins. Nonetheless, he dismisses CARDS findings as not meaningful due to this classification, despite admitting the study's relevance to women and its primary prevention status. Dr. Wells concedes that there are additional benefits for women with other risk factors but lacks specific knowledge about the levels of these risk factors, including LDL cholesterol. 

He contrasts this with his views on the CTT Collaboration study, which he critiques for improperly combining primary and secondary prevention results, despite it adjusting for cardiovascular risk. He argues that when assessing studies like CARDS or JUPITER, the risk level of participants is crucial, yet he simultaneously disregards studies that consider risk factors while merging primary and secondary prevention. This inconsistency undermines the reliability of his opinion under Rule 702. Furthermore, Dr. Wells critiques the Mora meta-analysis for including JUPITER, which he deems an outlier; however, JUPITER supports the effectiveness of statins in primary prevention for women. Ultimately, there is statistically significant evidence, including CARDS and JUPITER, indicating that statins are effective for primary prevention in women, irrespective of Dr. Wells's views on risk equivalents.

JUPITER and Mora provide statistically significant evidence regarding the effectiveness of Lipitor, which is subject to scrutiny under Rule 702. This rule mandates that expert testimony must be based on sufficient facts or data, allowing trial judges to assess the reliability of an expert's conclusions. In this case, the court finds a substantial analytical gap between the data presented and Dr. Wells' opinion, leading to its exclusion under Rule 702. 

Dr. Roberts contends that there is no convincing evidence supporting Lipitor's effectiveness for primary prevention of heart disease in women, noting that neither Lipitor nor any other statin has demonstrated a reduction in hard endpoints like heart attacks or strokes during primary prevention. She references the ASCOT and CARDS studies but refuses to consider CARDS evidence applicable to women, claiming that cardiovascular event reduction in the female subgroup was not statistically significant. However, this conclusion is criticized as the study was not adequately powered to detect differences in this subgroup, and the lack of significance might stem from insufficient power rather than an actual absence of effect.

Statistical tests for heterogeneity indicate that when gender heterogeneity is not significant, it is reasonable to infer that the treatment effects for women align with those of the overall population. In CARDS, these tests were not significant for sex, and treatment effect estimates remained consistent after adjusting for baseline factors. Despite Dr. Wells acknowledging beneficial effects of CARDS for women, Dr. Roberts neglects this evidence. The court cites precedent to emphasize that failing to consider contrary evidence undermines the reliability and scientific soundness of expert testimony, leading to the exclusion of Dr. Roberts' opinion.

The Court determined that Dr. Bérard's expert report selectively highlighted studies supporting her conclusions while neglecting contrary evidence, rendering her methodology unreliable and scientifically unsound. This issue was particularly evident in Dr. Roberts' assertion that Lipitor is ineffective for primary prevention in women, as she disregarded statistical evidence suggesting otherwise. Dr. Roberts also inadequately addressed evidence contrary to her views on statins, failing to mention relevant meta-analyses that contradict her conclusions, which undermines the reliability of her opinion. The Court cited prior cases emphasizing that an expert's opinion is unreliable if it does not acknowledge contrary evidence.

Similarly, Dr. Quon's analysis, which concluded there is no clinical benefit for women using Lipitor for primary prevention, was criticized for being based on a limited review of data. He focused solely on studies that supported his view, intentionally omitting significant studies that demonstrated positive outcomes. Specifically, while Dr. Quon referenced the Mora meta-analysis, he only discussed the portion that yielded non-significant results, misrepresenting the broader findings. This selective use of data further compromised the reliability of his opinion, leading the Court to exclude Dr. Roberts' efficacy opinions under Rule 702.

Dr. Hynes' testimony was excluded by the district court due to his selective use of facts, which does not adhere to the scientific method as required by Daubert standards. Similar issues were noted with Dr. Keegan, who also cherry-picked data to support his conclusions while disregarding contrary evidence, leading to the exclusion of his testimony. Dr. Quon’s opinion, stating no convincing evidence supports the clinical benefit of Lipitor for primary prevention in women, was similarly excluded under Rule 702 due to his failure to consider opposing evidence. Dr. Fleming did not provide an opinion on the efficacy of Lipitor for primary prevention but asserted that the ASCOT data did not demonstrate efficacy in women and that the drug’s label was misleading in this regard. However, claims about the misleading nature of the Lipitor label based on ASCOT are pre-empted, rendering Dr. Fleming’s testimony irrelevant and potentially confusing to the jury, resulting in its exclusion under Fed. R. Evid. 402 and 403. Dr. Abramson, in his report, opined that the ASCOT trial did not show benefits of Lipitor for women, supported by guidelines and meta-analyses indicating no evidence of statins providing primary prevention benefits to women. While he did not perform a comprehensive review, he stated in a deposition that there is no evidence supporting the efficacy of statins for primary prevention in women.

Dr. Abramson's opinion regarding the efficacy of statins for primary prevention in women lacks reliability because he has not addressed significant studies that support such efficacy, specifically the Mora meta-analyses and the Kostis study, which are published in peer-reviewed journals. His statement that there is "no evidence" to support this efficacy is deemed unreliable due to his failure to consider existing evidence. The Court emphasizes that an expert's opinion is unreliable if it ignores evidence that contradicts their theory. Additionally, Dr. Abramson exhibited reluctance to express this opinion outside of litigation, suggesting a disparity in the rigor of his expert testimony compared to his professional practice. While employed at Wells Fargo Insurance Services, he did not advise against reimbursement for Lipitor for primary prevention in women, despite testifying in litigation that it was ineffective. He acknowledged that his litigation work involves "boundaries" that do not apply to his regular professional engagements, raising concerns about his credibility. Consequently, the Court finds his opinion inadmissible under Rule 702 due to insufficient scientific rigor and reliability.

Dr. Abramson's deposition includes his opinion that the FDA should not have approved Lipitor for primary prevention in women, asserting it should be considered an off-label use. However, the plaintiffs do not argue against Lipitor’s approval for women generally, focusing instead on the appropriateness of the defendant's actions. Consequently, the court excludes Dr. Abramson's opinion under Federal Rules of Evidence 402 and 403. The court grants in part and denies in part the defendant’s motion to exclude expert testimony regarding Lipitor's efficacy for primary prevention in women. Claims based on the ASCOT data or diabetes risk related to Lipitor's labeling are preempted and dismissed. The court excludes Dr. Fleming's assertion that ASCOT data did not establish efficacy in women for primary prevention and also excludes Dr. Abramson's opinion on the FDA's approval. The plaintiffs do not contest Lipitor's effectiveness for secondary prevention in women and acknowledge its cholesterol-lowering properties. The court addresses the plaintiffs' claims regarding labeling changes and advertising for Lipitor, alongside the later FDA approvals for other indications, including primary prevention for diabetic patients based on the CARDS study. Pfizer does not argue that failure-to-warn claims regarding diabetes risk are preempted by federal law. The document notes the complexity of preemption analysis, especially given the diverse jurisdictions involved in the MDL.

Analyzing tort duties across 52 jurisdictions in relation to federal labeling requirements is deemed burdensome and unnecessary. Any state law that might compel the Defendant to alter statements about Lipitor's efficacy based on ASCOT data cannot be complied with alongside federal requirements. State tort law may be preempted if it imposes duties that conflict with federal law, as established in cases like Schrock v. Wyeth, Inc. and Drager v. PLIVA USA, Inc. Plaintiffs assert that Lipitor is ineffective for primary prevention of cardiovascular disease (CVD) in women without a history of coronary heart disease, while the Defendant's labeling states results for women are inconclusive. The FDCA preempts claims regarding misleading labeling, leading to the exclusion of related expert opinions. Dr. Roberts opines that the CARDS trial indicates no significant benefit of Lipitor for primary prevention in diabetic women, and critiques the JUPITER study's endpoints as inadequate. The 2004 indication for Lipitor includes reducing the risk of revascularization procedures and angina, which is relevant to the discussions around its efficacy. While Dr. Quon’s opinion of no benefit from Lipitor for women with diabetes aligns with previously addressed motions, the court permits consideration of his input. The court will separately address additional arguments concerning Dr. Abramson’s opinions regarding the JUPITER study.