Value of an in-depth analysis of unpublished data on the safety of influenza vaccines in pregnant women

Abstract

Background

Unpublished data can sometimes provide valuable information on the safety of biologic products.

Methods

We assessed information potentially available from regulatory authorities, manufacturers, and public health agencies. We explored 4 recently established vaccine registries, reviewed package inserts from 99 influenza vaccines, and contacted vaccine manufacturers and regulatory agencies for data on influenza vaccine safety in pregnant women.

Results

The vaccine registries did not have sufficient data to analyze and there are problems with the quality of the information. The majority of package inserts provided no product-specific safety information for pregnant women, especially in less developed countries. The majority of available data come from reports gathered from passive adverse event reporting systems in the general population and reports of women enrolled in clinical trials of influenza vaccines who became pregnant at various times before or after receiving influenza vaccine. The information was not collected in a systematic manner, there are inconsistencies in the follow up of pregnant women and the available information about pregnancy outcomes. Considerable resources would be needed to systematically identify all of the information, try to obtain missing follow up information, and conduct analyses. There would be substantial limitations to any attempt to conduct a systematic analysis.

Conclusions

The value of trying to analyze unpublished data on the safety of influenza vaccine in pregnancy is limited and would require considerable resources to thoroughly investigate. Expanding efforts to identify and review unpublished data regarding the safety of influenza vaccines in pregnancy is not likely to produce information of high scientific value or information that could not be identified from publications and other publically available data.

1. Introduction

The World Health Organization’s (WHO) Strategic Advisory Group of Experts (SAGE) has recommended maternal influenza immunization [1]. As part of the overall effort WHO requested a review of the potential value of conducting an in-depth analysis of unpublished data on the safety of influenza vaccine during pregnancy.

2. Methods

We conducted online searches to identify public datasets and accessed potential sources of unpublished data from vaccine manufacturers, online package inserts, regulatory agencies, and online clinical trials databases for information relevant to the safety of influenza vaccines during pregnancy (Table 1). We also contacted some key individuals responsible for this information for assistance.

Table 1.

Sources and URLs.

Organization		Source	URL/Website
	Collaboration: Astellas, Bayer, Boehringer Ingelheim, Daiichi Sankyo, Eisai, GSK, Lilly, Novartis, Roche, Sanofi, Takeda, UCB and ViiV Healthcare	Clinical Study Data Request	clinicalstudydatarequest.com
EMA	European Medicines Agency		www.ema.europa.eu/ema
EU	European Union	EudraCT – EU Clinical Trails Register	www.clinicaltrialsregister.eu/ctr-search/search
FDA	(US) Food and Drug Administration	Vaccines Licensed for Use in the United States	www.fda.gov/BiologicsBloodVaccines/Vaccines/ApprovedProducts/ucm093833.htm
FDA	(US) Food and Drug Administration	Postmarketing Requirements	www.accessdata.fda.gov/scripts/cder/pmc/index.cfm
FDA/CDC	(US) Food and Drug Administration and (US) Centers for Disease Control and Prevention	VAERS – Vaccine Adverse Event Reporting System	vaers.hhs.gov/
NIH	(US) National Institutes of Health	Clinical Trials	www.clinicaltrials.gov
WHO	World Health Organization	International Clinical Trials Registry Platform	apps.who.int/trialsearch

The European Medicines Agency (EMA) provided information to us dated May 19, 2015 about all the manufacturers that had obtained central approval via the EMA. The Documents Manager conducted several searches for us and abstracted the data available (Appendix 1). There was no specific data provided in the database they sent, but study numbers within the database could be used to conduct a search on the EMA site to find specific study data and results. We also searched for information on the EMA website.

The US Food and Drug Administration (FDA) maintains a website with information on licensed vaccines. The information available on the section “Vaccines Licensed for Use in the United States” lists product names and trade names and links to documentation regarding product approval, new indications and package inserts. A manufacturer submitting a Biologic License Application (after finishing the 3 phases of clinical development) must submit safety and efficacy data available for the product, including published and unpublished studies. A summary of postmarketing requirements is available online.

Vaccine manufacturers: In 2015, we identified 47 manufacturers of 91 influenza vaccines distributed in the EU and the US, Canada, Australia, New Zealand, Taiwan, India, China with information available online [2], this information was updated to include more recent information on a total of 99 influenza vaccines. All 99 of these package labels/inserts were available online and were reviewed for relevant information.

2.1. Clinical trials data

Clinicaltrials.gov is a registry and database of clinical studies conducted in human subjects throughout the world (190 countries) maintained by the US National Institutes of Health. Conducting an Advanced Search using Other Terms: ‘influenza vaccine NOT Haemophilus NOT Hib'’, Conditions: ‘pregnant OR pregnancy OR maternal’.

EU Clinical Trials Register (EudraCT) is a database of clinical trials within the European Economic Area (EEA). Trials with sites outside the EEA are included if they are “marketing authorisation holder-sponsored and involve the use in the paediatric population of a medicinal product covered by an EU marketing authorisation … or if they form part of an agreed PIP (Paediatric Investigation Plan)”. This registry includes pediatric clinical trials and Phase II-IV adult clinical trials.

WHO International Clinical Trials Registry Platform (ICTRP) is a registry containing ongoing and completed clinical trials. This searchable site contains trial data sets around the world. The ICTRP was searched for trials using Advanced Search with the following parameters – Title: “influenza vaccine” AND Condition: “pregnant OR pregnancy OR maternal”, NOT Intervention: “Haemophilus” and setting Recruitment status to ALL (the default is “Recruiting”).

Clinical Study Data Request includes studies from sponsors who have stated commitments to use this open access site. A search of the Clinical Study Data Request site was conducted in July 2017 with the following parameters: Search all Sponsors, Find by: Medicine = Influenza Vaccine (there was an option for Medical Condition but nothing to indicate pregnancy). This site did not allow for an Advanced Search.

The US Centers for Disease Control and Prevention (CDC): The Immunization Safety Office coordinates assessments of individual adverse events, post-licensure clinical trials, and epidemiologic investigations of safety for vaccines marketed in the United States. Unpublished data from these investigations are sometimes presented at meetings of the Advisory Committee on Immunization Practices (ACIP) which makes recommendations for the use of vaccines in the United States. We monitored ACIP meetings via webcast and searched the CDC website for additional information. Investigators at the CDC periodically present unpublished data at ACIP meetings.

2.2. Vaccine Adverse Event Reporting System (VAERS) and other passive reporting systems

The CDC and the FDA maintain the Vaccine Adverse Events Reporting System (VAERS) a national post-marketing vaccine safety surveillance program of reported adverse events following immunizations administered primarily in the United States. Reports are accepted for vaccines administered in other countries, but analyses are done primarily for data obtained from the United States. Other countries and the European Union have separate databases similar to VAERS.

3. Results

The EMA requires manufacturers to conduct post licensure safety studies and requires manufacturers to provide risk management plans (RMPs) to address theoretical concerns about safety in pregnant women for some influenza vaccines. Consulting with the Documents Manager or searching through their database was time-consuming and inefficient.

A search of the FDA’s database on the Postmarket Requirements and Commitments page on August 3, 2017 (www.accessdata.fda.gov/scripts/cder/pmc/index.cfm) using only the parameter “Product: influenza vaccine” resulted in 13 Applications/Supplements. Of the 6 applications relevant to pregnancy registries 3 are ongoing and 3 are pending; none had data posted.

Some of the influenza vaccine manufacturers with the highest sales revenue in 2015 are listed in (Table 2). The package labeling/inserts provided by manufactures for most influenza vaccines contains wording about use of the product in pregnant women; however, much of the information is vague, cites animal studies or general information not specific to the product. The wording in the package inserts regarding use of the vaccines in pregnancy varies considerably ranging from “…no evidence of harm…” to “…forbidden to use…” As noted by Regan [3], this information contradicts the clear recommendations for the use of influenza vaccine in pregnant women and adds to confusion of providers and consequently leads to lower rates of maternal vaccination.

Table 2.

Influenza vaccine manufacturers, vaccines, sales revenue and package insert wording.

Company	Vaccine	Estimated sales	Package insert information on pregnancy	PI Year
Abbott	Influvac	$188 million	Inactivated influenza vaccines can be used in all stages of pregnancy. Larger datasets on safety are available for the second and third trimester, compared with the first trimester; however, data from worldwide use of influenza vaccine do not indicate any adverse foetal and maternal outcomes attributable to the vaccine	2014/15
AstraZeneca	FluMist	$162 million	… no evidence of impaired fertility or harm to the fetus due to FluMist Quadrivalent. There are however, no adequate and well controlled studies in pregnant women. Because animal studies are not always predictive of human response FluMist Quadrivalent should be administered during pregnancy only if clearly needed	2014/15
GlaxoSmithKline	FluLaval/Fluviral	$375 million	…no evidence of impaired female fertility or harm to the fetus due to FLULAVAL QUADRIVALENT. There are, however, no adequate and well-controlled studies in pregnant women. Because animal reproduction studies are not always predictive of human response, FLULAVAL QUADRIVALENT should be given to a pregnant woman only if clearly needed	2014/15
Johnson & Johnson	Inflexal V	$35 million	Limited data from flu vaccinations in pregnant women do not indicate that the vaccine would have harmful effects on the pregnancy or the baby. The use of this vaccine may be considered from the second trimester of pregnancy. For pregnant women with medical conditions that increase their risk of complications from the flu, administration of the vaccine is recommended, irrespective of their stage of pregnancy	2011/12
Laboratorios Farmacéuticos ROVI	Levrison	$3 million	No info found
Mitsubishi Tanabe Pharma	BIKEN HA	$108 million	No info found
Novartis	OptaFlu	$71 million &	In general data from influenza vaccinations in pregnant women do not indicate adverse foetal and maternal outcomes attributable to the vaccine. Animal studies do not indicate reproductive toxicity. The use of Optaflu may be considered from the second trimester of pregnancy. For pregnant women with medical conditions that increase their risk of complications from influenza, administration of the vaccine is recommended, irrespective of their stage of pregnancy	2014/15
Novartis	Fluvirin	$359 million	No evidence of impaired fertility or harm to the fetus due to FLUVIRIN®. There are, however, no adequate and well-controlled studies in pregnant women. Because animal reproduction studies are not always predictive of human response, this vaccine should be used during pregnancy only if clearly needed	2014/15
Sanofi	Fluzone (sold as Vaxigrip outside the US)	$1.343 billion	Animal reproduction studies have not been conducted with Fluzone. It is also not known whether Fluzone can cause fetal harm when administered to a pregnant woman or can affect reproduction capacity. Fluzone should be given to a pregnant woman only if clearly needed	2014/15
Sinovac Giotech	Anflu	$10 million	This product is forbidden to use for the groups below: (3) Women during pregnancy	2010

Seasonal influenza vaccine was recommended for use during pregnancy in 29% of the identified package inserts; 44% recommended use with a caveat (ie, ‘if clearly needed’, ‘if benefits outweigh risks’) (Table 3). Only 2 inserts for seasonal influenza had language stating that the vaccine was not recommended.

Table 3.

Pregnancy and breastfeeding information in 99 influenza vaccine package inserts available online.

	Seasonal	Pandemic	Total
Pregnancy information	68	31	99
Can be used	20	1	21
Not recommended or forbidden	2	1	3
Caveat for use	30	28	58
No recommendation	7	1	8
No Package Insert available online	6	0	6
Not applicable due to age group	3	0	3
Breastfeeding information	68	31	99
Can be used	30	12	42
Not recommended	1	1	2
Caveat for use	16	15	31
No Package Insert available online	6	0	6
No recommendation	11	3	14
Not applicable due to age group	4	0	4

In the US FDA’s most recent relevant regulations, effective June 30, 2015, package inserts must remove the old Pregnancy Categories A, B, C, D and X. Labels must now contain information about Pregnancy Registries and a summary of the risks of use during pregnancy and lactation, a discussion of the supporting data, and relevant information to help health care providers make decisions and counsel women about use during pregnancy and lactation. Depending on original approval dates, the changes to the labels must be submitted for approval by the end of June 2018–2020.

For the limited number of package inserts that have incorporated the new information, the recommendations are even more limited than before the change. A statement indicating that there are insufficient data regarding administration to pregnant women is typically included and no recommendations are provided.

Currently, most package inserts indicate that there are no available data from clinical trials involving pregnant women. Some manufacturers provide limited information about women who became pregnant after being enrolled in trials that exclude pregnant women. There are some inserts that provide results of trials where placebo or comparison vaccines were given. Identifying more such trials and comparing women in both study arms who inadvertently became pregnant would only provide small numbers and it is unlikely that meaningful information could be obtained because of the variability in types of vaccine, enrollment criteria, timing of the pregnancy, and other factors.

3.1. Registries

Some manufacturers have maintained registries of women who received influenza vaccines during pregnancy or in the few weeks prior to pregnancy. We contacted 3 manufacturers requesting information from their pregnancy registries (from information in package inserts): Novartis (manufacturer of Flucelvax), Sanofi (manufacturer of FluQuadri in Australia and New Zealand) and GlaxoSmithKline (manufacturer of Fluarix/Quadrivalent and FluLaval/Quadrivalent). Novartis provided results of animal studies that had already been obtained from the package insert. The Flucelvax Pregnancy Registry (Novartis) is currently recruiting participants and is estimated to be completed in June 2020.

For other registries, the information available usually includes only summary information regarding the numbers of women who became pregnant, many of whom became pregnant weeks or months after the vaccine was given. For women immunized while pregnant there is an intent to follow-up the pregnancies, but the information available is often limited.

In a setting where there is a recommendation for all pregnant women to receive influenza vaccine, selection bias is likely to occur for reports to registries. Careful review of all reports would be needed to document reporting to the registry prior to pregnancy outcomes and prospective followup to obtain the outcomes. No such information was available in the limited data we reviewed.

The EMA provided additional information regarding other manufacturers’ commitments to establishing registries, but no information of substance has been identified that would provide previously unidentified data to inform on the safety of influenza vaccines in pregnancy.

3.2. Clinical trials

A search of clinicaltrials.gov as described above resulted in 62 studies. Limiting this search to ‘Study Results: With Results’, 7 completed trials with results were identified. One trial was testing the effect of text message reminders on uptake of influenza vaccine among pregnant women and was not relevant to safety issues.

A search of the EudraCT using search terms: ‘(influenza vaccine NOT Haemophilus NOT Hib) AND (pregnancy OR pregnant OR maternal)’ on July 20, 2017 produced 119 results. Using the Advanced Search with the same terms, Select Date Range for ‘2011 to 2017-07-20’ and setting Results Status to ‘Trials with results’ produced 28 results, all were trials that excluded pregnant women.

Searching WHO’s ICTRP, 6 studies were identified. Removing NOT Intervention: “Haemophilus” resulted in 9 trials, but none had posted study results. One study had an objective to examine the effect of education on vaccine uptake in pregnancy and was not relevant to safety data. Four studies have not yet reached the completion dates and four other studies were complete but did not have results posted.

In a search of the Clinical Study Data Request, selecting ‘influenza vaccine’ from the pulldown menu produced 134 studies; the same number as typing ‘influenza vaccine’ into the search box to search by study number or study title. There was no obvious way to further narrow the search.

Data from CDC’s Immunization Safety Office are almost always eventually made public, but it is possible that some of data may not be published. Since most studies presented to the ACIP are eventually published, investing considerable effort to find the data may not be an optimal use of resources. Inquiries to CDC or any other U.S. federal agency will usually provide answers regarding the data.

3.3. VAERS

There are many limitations to the evaluation of passive reports and there is a general consensus that causal relationships can usually not be determined based upon individual reports. Limitations include underreporting, selective reporting of adverse events, lack of accurate denominator data, incomplete data on adverse events, and lack of specific evidence for the diagnosis submitted in many reports [4]. However, analyses of reports may provide a signal for potentially previously unrecognized adverse events, or higher than expected rates of adverse events that might warrant further epidemiological and clinical investigations.

The VAERS data set is periodically analyzed by personnel from CDC and FDA. A 2014 analysis of the safety of live attenuated influenza vaccine administered to pregnant women did not reveal any suggestion of increased rates of complications associated with pregnancy although there was a disproportionate number of cases of Guillain-Barré syndrome reported [5].

CDC and FDA investigators have published reviews of VAERS reports of influenza vaccines in pregnant women [6], [7], [8] and they have conducted additional analyses of VAERS data for the years 2010–2015 [9]. Preliminary data were presented at the 31st International Conference on Pharmacoepidemiology, August 23–26, 2015 in Boston, MA [10] and a study of reports for 2010–2016 was published in 2017 revealing no new or unexpected patterns in maternal or fetal outcomes in their report of VAERS data from 2010-2016 [11]. The investigators at these agencies are familiar with problems in analyzing and interpreting these passive reports therefore, there is no need for any further attempts to request or analyze unpublished data submitted to VAERS.

4. Discussion

4.1. Value

The strongest evidence for assessing vaccine safety comes from randomized controlled clinical trials (IOM 2012). Websites that provide valuable information about ongoing and completed clinical trials include WHO-ICTRP, EU-CTR, ClinicalTrials.gov and Clinical Study Data Registry. All the results from searching each of these sites include the identifier code provided by ClinicalTrials.gov which begins with the letters ‘NCT’. To date, we found it most advantageous to go directly to ClinicalTrials.gov as none of the other sites provided information that was not also found on ClinicalTrials.gov, and ClinicalTrials.gov was the site that provided the most data. Most clinical trials targeted at answering questions about influenza vaccine in pregnancy appeared to be published and recently conducted and ongoing studies are likely to be published. We did not identify any older trials that enrolled moderate to large numbers of pregnant women who received influenza vaccine that were not published. We assume that the ongoing studies identified have a high likelihood of being submitted for publication.

Most randomized controlled clinical trials require large amounts of resources and the results can require several years to be fully analyzed and published. For example, a recently published trial of maternal influenza immunization in Nepal was published almost 4 years after the last woman was enrolled [12]. Identifying data from unpublished completed trials does not mean that the data will not be published and an analysis of data by individuals not familiar with the complexities of the trial and analyses could lead to faulty interpretations. Clinical trials in pregnant women are usually underpowered to identify small increases in risk of adverse pregnancy outcomes associated with vaccines [12].

Large-scale post licensure epidemiologic studies of pregnant women who received vaccines compared to controls who received other vaccines or were unvaccinated can provide high quality scientific evidence to evaluate pregnancy outcomes and adverse events that are too rare to be detected in prospective randomized trials. For example a study using the Vaccine Safety Datalink compared 52,856 infants with maternal first trimester IIV exposure to 373,088 infants whose mothers were unexposed and found no evidence in an increase in major structural birth defects associated with maternal influenza vaccination [13]. Also, receipt of influenza vaccine during pregnancy was associated with age, underlying conditions, and ethnicity, factors which could be confounding for some pregnancy outcomes in uncontrolled studies. Other biases and confounding associated with influenza vaccination during pregnancy have been found including variable access to vaccines [14]. Analysis of the databases requires special expertise. Nevertheless, data from these studies can be useful for identifying signals of possible problems and ruling out evidence of hypothesized adverse outcomes associated with vaccines [15], [16].

There is no systematic database for unpublished studies so special efforts would be needed to identify unpublished data and finding appropriate controls would be very difficult, if not impossible.

There appears to be limited information that could be obtained from the incidental reports of pregnant women who were inadvertently included in clinical trials that were intended to exclude pregnant women. It is unlikely that significant new findings would be obtained because the numbers are small, follow-up information is often incomplete, the timing of the vaccine with regard to pregnancy is not controlled, and comparison data with unvaccinated women in the same population is usually unavailable.

Pregnancy registries could possibly provide clues as to previously unanticipated adverse events, but they would be unlikely to provide any valuable information about increased rates of common adverse events that occurred during pregnancy such as spontaneous abortions, prematurity, preeclampsia, etc. Controlled trials are needed to determine if there are differences in rates of these adverse events and women who received influenza vaccine versus controls.

Passive reports of adverse events following immunizations can provide a signal of a possible problem, but there are many limitations to such data and causal relationships cannot usually be determined from passive reports [17]. Formal systems to analyze these passive reports are in place in many highly industrialized countries, and efforts are being made to implement systems in countries throughout the world [18], [19]. These analyses can serve as a signal for new and previously unrecognized adverse events as well as unexpected higher rates of interest events with one vaccine as compared to others. Reports made to manufacturers in most industrialized countries are forwarded to public health agencies who have conducted some analyses. Investigators working with systems in industrialized countries have indicated that there will be continued monitoring and publication of results. Therefore, we would not recommend any additional effort be pursued to investigate these reports at this time. We did not investigate the availability of unpublished data in less-developed countries as this would be far beyond the scope of this work.

4.2. Feasibility

Unpublished data on the safety of influenza vaccines in pregnancy can be identified through online sources. Assistance can be obtained from key personnel familiar with these systems at regulatory agencies and possibly other sources. However, unpublished data collected prior to the establishment of required reporting of clinical trials may not be identified. Some manufacturers have committed to making new data available on online websites soon after the results are available. It is difficult to determine if currently unpublished data will be submitted for publication, or has already been submitted and undergoing review. For clinical trials, such data are now readily available and the expectation is that recently initiated and future clinical trials will probably be routinely submitted for publication. Ongoing population-based epidemiologic studies are not registered and so it is difficult to identify unpublished data, but it might be possible to obtain information from organizations that have conducted such studies in the past. Other sources of unpublished data are not likely to provide strong evidence with regard to assessing calls of relationships between adverse events and influenza vaccines administered in pregnancy.

There are differences between types of studies and databases available. Multiple protocols to obtain, analyze, and review data would be required. A preliminary review indicates that an attempt to make a single generic protocol to review all types of databases would not be possible because of the differences in studies and reporting. Also, there is no standardization to the formats for the data from different studies. Attempts to consolidate the data into a format that would allow for analyses would take an enormous effort which might not be successful. Separate protocols for reanalysis of data from each study may be needed. Also, there are differences in the manufacturing processes and safety profiles of influenza vaccines [1]. Combining information from multiple different vaccines produced by different methods may not be appropriate.

Active surveillance of pregnancy outcomes in women who received vaccines and compared with women who had not received the vaccines are subject to selection bias and the healthy vaccine effect, but these studies have been very useful for identifying or ruling out possible signals of problems associated with the use of vaccines in pregnancy [15], [16].

5. Conclusions

Conducting a detailed review of unpublished data appears to be impractical and would require considerable resources. We do not believe that the results would contribute additional valuable information with regard to the safety of influenza vaccine in pregnancy. A more appropriate use of resources would be to invest in additional active surveillance studies of vaccines in pregnant women.