1 Introduction

HPV infection is a necessary step in the pathogenesis of cervical, other anogenital and some non-genital cancers [1, 2]. Primary prevention of infection with oncogenic HPV types has the potential to prevent morbidity and mortality worldwide. Cervical cancer alone is the fourth most common cause of cancer-related death worldwide [3, 4]. In addition to causing cancer, HPV infection also causes genital warts, the most common sexually transmitted disease in many developed country settings [57].

Two HPV vaccines are available: the bivalent vaccine Cervarix® (2vHPV, GlaxoSmithKline Biologicals, Belgium), which protects against the two oncogenic HPV types 16 and 18 that account for ~70 % of cervical cancer, and the quadrivalent vaccine Gardasil® (4vHPV, Merck and Co., USA), which protects against 16 and 18, as well as the non-oncogenic types 6 and 11, predominantly responsible for genital warts and other non-malignant lesions. HPV antigens in both vaccines are composed of L1 proteins specific to each HPV type, which are derived using recombinant technology (yeast or insect cell in vitro expression systems) and form conformationally intact non-infectious virus like proteins (VLPs). The vaccines also contain adjuvants, which assist in enhancing the humoral immune response. For 4vHPV the adjuvant is a proprietary aluminium hydroxyphosphate sulphate system and for 2vHPV the adjuvant system is called AS04 and contains both an aluminium salt and monophosphoryl lipid A. [8, 9].

The clinical efficacy of these vaccines in the prevention of persistent HPV infection and intraepithelial neoplasia (potentially pre-cancerous lesions) at the cervix in women has been demonstrated in pre- and post-licensure studies [1012]. Following impressive clinical trial results in females, both vaccines were made available for use from 2006–2007, with the 4vHPV vaccine now registered in 127 countries and an estimated >95 million doses distributed worldwide (Dr Carlos Sattler, personal communication, Merck, July 2012) and 2vHPV registered in >115 countries with >33 million doses distributed (Kristin Verschueren, personal communication, GSK, July 2012). Recently, population-based vaccination of males using 4vHPV has also been recommended in countries such as the USA, Canada and Australia [1316] based on efficacy of this vaccine against vaccine-type persistent infection and intraepithelial neoplasia in anogenital sites in males [17]. The widespread availability of these vaccines with expanding use into new settings, particularly less developed countries, underpins the importance of reviewing the body of evidence for safety. Although there is no a priori reason to expect that the safety of HPV vaccines would be particularly different to other inactivated vaccines[18], concerns regarding perceived HPV vaccine safety issues have at times received extensive media attention and have the potential to reduce vaccine uptake [19, 20]. This review aimed to assess all available published safety data on both HPV vaccines, including randomised clinical trials, meta-analyses and data from post-licensure studies.

2 Methods

Articles cited in this review were obtained by searching OVID Medline (1946–May 2012) and OVID EMBASE (1980–May 2012) databases up to May 2012. Individual searches were completed on HPV vaccine safety, HPV vaccine trials and post-licensure studies, and HPV vaccines and pregnancy. Both database controlled vocabulary terms and commonly used free-text terms for HPV vaccines and/or safety were used: a full list of all search terms is available on request. There was no language restriction; however, only articles with English language (abstract or more) were reviewed. The bibliographies of identified articles and reviews were hand-searched to identify additional studies. Internet search engines were also queried for ‘HPV vaccine case reports’. Formal inclusion and exclusion criteria were not applied for the purposes of selecting studies for the review; however, the authors endeavored to cite all studies (or pooled analyses) that contained original data, irrespective of the study type (e.g. controlled trial, case report, etc). Position statements and immunisation guidelines were also reviewed.

3 Pre-Licensure Controlled Clinical Trials and Long-Term Follow-Up Studies

The pre-licensure studies examining the safety of both 4vHPV and 2vHPV vaccines were extensive. In addition, a number of controlled clinical studies have continued past their pre-specified completion dates, with long-term follow-up of safety as well as immunogenicity/efficacy outcomes. Additional randomised controlled studies have been conducted in new populations in Korea, China, Japan and Vietnam [2125]. Safety endpoints in most studies included local and systemic adverse events (AEs), serious AEs (SAEs), death and new onset medical conditions, including chronic and/or autoimmune disease. Pregnancy outcomes were also assessed and are discussed separately below. Many results presented below are from pooled analyses, which give the advantage of including large numbers of participants. Despite ongoing efforts to harmonise the categorisation and reporting of vaccine safety outcome measures in clinical trials [26], limitations on combining results from studies with different designs and/or outcome measures exist. Nevertheless, the results for most studies within these analyses were generally consistent.

3.1 Summary for 4vHPV from Clinical Trials

A pooled analysis on several early and pivotal trials [2731] involving a total of >20,000 females aged 9–26 years and about 1,350 males aged 9–16 years, mostly from Europe, North and Latin America, showed that injection site reactions (ISR) were significantly more common in vaccine recipients compared with recipients of aluminium-containing placebo or non- aluminium placebo injections [32]. Injection site pain was most common (83 vs. 77 vs. 49 %, respectively, for the 3 recipient groups) with severe pain reported in 4 % of recipients compared with 2 % receiving aluminium-containing placebo. Erythema (24 versus 18 or 13 %) and swelling (24 versus 16 or 8 %) were also more common in vaccine recipients. Common systemic adverse experiences did not differ markedly between groups, with headache most common (26 %), followed by fever (13 %) and nausea (6 %) [32].

In this analysis there was no significant difference in the frequency of SAEs overall or by system organ class (9 groups) over a median follow-up period of 3.6 years and >34,000 person-years-at-risk [32]. The five reported vaccine-related SAEs were not of any particular cluster [32]. Deaths occurred in 0.1 % of both vaccine and control recipients, respectively, with no death deemed related to the vaccine study [32]. The overall proportion reporting new onset autoimmune conditions was not different between both groups (2.4 % in each); specific conditions that were nominally higher among 4vHPV vaccine recipients included thyroiditis, rheumatoid arthritis and proteinuria, although each occurred in <0.1 % [32]. However, given the relative rarity of these individual conditions, even analysis of pooled data from many clinical trials does not provide enough participants to detect meaningful differences in such low-incidence events (discussed further in Sect. 5).

Another pooled analysis of trials in approximately 6,000 Latin American females aged 9–24 years also found more ISR in vaccine compared with placebo recipients (85 versus 73 %). Proportions with any systemic reaction (~60 %) and SAEs (<0.5 %) were comparable between vaccine and placebo recipients [33]. Smaller studies in females from Mexico and Korea reported similar frequencies of local and systemic AE [24, 34]. An open-label study in Vietnam comparing four 4vHPV vaccine dosing schedules found slightly lower rates of injection site pain (<69 % for any vaccine dose/schedule) [22]. With respect to age, in one trial a significantly lower proportion of younger (10–15 years) compared with older (16–23 years) females reported injections site pain (79 versus 86 %) and redness (20 versus 26 %). However, more younger females reported fever (12 versus 7 %) [27]. In a study of 4,000 older women (24–45 years) safety was comparable to younger females, with modestly lower frequencies of both local and systemic reactions [35, 36].

The 4vHPV vaccine was evaluated in a head-to-head comparison study between boys and girls aged 10–15 years in which local and systemic events were comparable between genders [27]. In one study of 4,000 males aged 16–26 years, ISRs were significantly more common among 4vHPV vaccine compared with aluminium-containing placebo recipients (60 versus 54 %) [37]; however, severe pain was uncommon in either group [17]. These frequencies were lower than reported among females in trials of similar design. Reporting of SAEs and new medical conditions was comparable between male vaccine and placebo recipients [17, 37], and safety in men who have sex with men, a subset of all males in these studies, was consistent with that for the whole study population [37]. Among subjects seropositive at baseline for at least one vaccine HPV type, the proportions reporting injection site reactions were similar to that of all subjects, and in both females and males [28, 29, 37].

When specifically reported, the frequency and severity of AEs following subsequent 4vHPV vaccine doses appeared similar, if not reduced, compared with AEs after the first dose [28, 29]. Two small studies in young females noted a decreasing proportion of recipients reporting injection site pain but increasing proportions reporting erythema and swelling with subsequent doses [27, 30]. Another study reported fewer participants with fever after each dose (5.1, 4.3 and 2.6 % following doses 1, 2 and 3 respectively) [24]. In males there was no increase, and suggestion of some decrease, in ISR and systemic adverse events with successive doses [37]. One small study reported a small increase in reactogenicity when more than three doses of vaccine were given; adverse events occurred in 80 % of women after a fourth dose at month 60 compared with 65.5 % after dose 3 at month 6. However, the majority of AEs were mild to moderate [38].

Studies of concomitant vaccination have generally demonstrated no or little change in the AE profile for the 4vHPV vaccine [30, 3941]. For example, adverse reactions in adolescents given the 4vHPV vaccine concurrently with the hepatitis B vaccine, or alone, were comparable [41]. A modestly higher proportion of adolescents (4 %) reported ISR from 4vHPV and some systemic reactions (8 % more with headache) after concurrent administration of 4vHPV vaccine with the diphtheria-tetanus-acellular pertussis-inactivated poliomyelitis (dTpa-IPV) vaccine [40].

3.2 Summary for 2vHPV from clinical trials

The safety profile from clinical studies of the 2vHPV vaccine appears generally consistent with that for 4vHPV vaccine. A pooled analysis of 11 studies in about 30,000 females aged ≥10 years (with approximately 16,000 receiving at least one dose of 2vHPV and almost 46,000 vaccine doses administered) showed a higher incidence of ISR (pain, redness, swelling) in 2vHPV vaccine recipients compared with controls given an aluminium-containing control vaccine or hepatitis A vaccine [42]. The published trials within this review reported similar safety outcomes [10, 21, 25, 4353]. Pain was the most common symptom (approximately 80 %) in all age groups and occurred in up to 97 % of adolescent girls [51]. Severe pain was reported in up to 7.5 % of recipients aged 15–25 years and severe ISR was more common in vaccine recipients (up to 0.6 % for redness and 1.2 % for swelling versus 0.1 and 0.2 %, respectively) [42]. However, ISRs were transient, generally lasting <5 days and mostly 2–3 days [10, 21, 25, 4353].

Up to 55 % of 2vHPV vaccine recipients reported systemic symptoms, most commonly fatigue, headache and myalgia. Selected systemic symptoms were more common in vaccine recipients compared with control recipients in some, but not all, studies. There were no significant differences in SAEs, unsolicited symptoms and medically significant conditions [42]. Five deaths were reported (1 in a 2vHPV vaccine recipient and 4 in control recipients). However, no deaths were considered related to vaccination [42]. In follow-up studies, over 4 or more years, rates of vaccine-related SAEs, new-onset chronic disease and new-onset autoimmune disease were no different between groups [10, 44, 54, 55]. In addition, the proportions reporting significant medical events were similar among groups when stratified by age [56].

The safety profile of 2vHPV from smaller studies in new settings/populations has been consistent with earlier large studies, albeit with some limitations due to small numbers of participants. These have included studies in Korean girls (aged 10–14 years) [25] and women[46], a phase I study in 30 females in China [23] and a study in 50 girls aged 9–13 years in Bangladesh [45].

The frequency of ISR following 2vHPV vaccination, as well as systemic symptoms, appeared to diminish with increasing age among females [51], although in one study the AE profile was similar in those aged 10–14 and 15–25 years, respectively [49]. One study compared the three-dose schedule of the 2vHPV vaccine with two-dose schedules using the standard or double-strength formulations of the 2vHPV vaccine among women aged 9–25 years [57]. The AE profile over a 2-year follow-up period did not differ between the four groups, including those receiving the higher antigen content vaccine doses [57]. There is only one published trial of the use of 2vHPV vaccine in males and this did not include direct comparison with females. However, the AE profile reported in males was consistent with that from female studies [50].

Rates of AEs following concurrent administration of the 2vHPV vaccine with other vaccines (diphtheria-tetanus-acellular pertussis-inactivated poliomyelitis vaccine, diphtheria-tetanus-acellular pertussis vaccine, meningococcal conjugate vaccines, hepatitis B vaccine and the combined hepatitis A/hepatitis B vaccine) were generally comparable to those of participants given the 2vHPV vaccine alone [5861] and similar to those for the 2vHPV vaccine administered alone in other studies. There was no increase in the frequency of solicited local or systemic adverse events with subsequent doses in the three-dose course of the 2vHPV vaccine, irrespective of whether the vaccine was administered alone [42, 48, 49] or concurrently with other vaccines [58, 61].

The 2vHPV vaccine contains a unique adjuvant called ASO4. A study investigating autoimmune diseases in all individuals enrolled in randomised, controlled trials of ASO4-containing vaccines (n = 68,512) over a mean follow-up period of 21 months found an overall rate of around 0.5 % for autoimmune events that did not differ between the AS04 and control groups [62]. The relative risk of an autoimmune event was 0.92 (95 % CI 0.70, 1.22) for the HPV 16/18 vaccine specifically.

3.3 Meta-Analyses of Clinical Trials

A number of meta-analyses of safety data from clinical trials of HPV vaccines have been conducted. One study [63] included data from six randomised controlled trials conducted up to June 2007 involving 2vHPV [44, 48, 52] and 4vHPV vaccines [12, 28, 29, 31], as well as the prototype monovalent type 16 HPV L1-VLP vaccine [64] in comparison with various controls. In the 40,323 female participants aged approximately 15–25 years, the incidence of SAEs in the vaccine and control groups was similar (odds ratio 0.998, 95 % CI 0.87–1.14). From four major studies that reported on death following HPV vaccine administration (two each for 4vHPV and 2vHPV), a total of ten deaths occurred in vaccine recipients and 11 deaths in control recipients (odds ratio 0.91, 95 % CI 0.39–2.14). No deaths were considered attributable to vaccination [63]. A more recent meta-analysis [65] that pooled data from seven unique clinical trials [10, 12, 28, 29, 31, 36, 44, 48, 52, 64, 6668] involving 44,142 females given either 2vHPV or 4vHPV (or control/placebo) vaccines found no increase in the risk of SAEs among vaccine recipients (risk ratio 1.00; 95 % CI 0.91–1.09). There was a trend, which did not reach statistical significance, toward an increased risk of injection site-related SAEs among vaccine compared with control recipients (risk ratio 1.82; 95 % CI 0.79–4.20) [65].

3.4 Comparing 2vHPV Vaccine and 4vHPV Vaccine Based on Clinical Trials

One observer-blinded randomised head-to-head study compared 2vHPV and 4vHPV vaccines directly [69]. In the approximately 1,100 women aged 18–45 years, solicited AEs were significantly more common in 2vHPV compared with 4vHPV vaccine recipients: 95.1 % (95 % CI 92.8–96.7 %) versus 85.1 % (95 % CI 81.8–88.1 %), respectively [69]. Injection site pain, redness and swelling, and fatigue and myalgia were more common in 2vHPV vaccine recipients, as was severe ISR [17.4 % (95 % CI 14.2–20.9 % for 2vHPV) and 3.4 % (95 % CI 2.0–5.4 %) for 4vHPV, respectively]. However, overall most AEs were transient and of mild or moderate severity. At 24-month follow-up, the proportions reporting SAEs, significant medical conditions, new onset chronic diseases or autoimmune diseases were similar between the two groups [70]. Of note, and possibly related to the propensity for more ISR following 2vHPV, the magnitude of the immune responses to each vaccine across all age strata was generally greater for 2vHPV. For example, geometric mean titres of serum neutralizing antibodies were 2.3–4.8-fold higher for HPV-16 and 6.8–9.1-fold higher for HPV-18 after 2vHPV compared with 4vHPV [69]. The higher ISR rates among 2vHPV vaccine recipients seen in this study needs to be considered against this evidence for better immunogenicity, although the clinical significance of the latter is not known. The safety results of this single head-to-head study are generally consistent with that from separate studies of both vaccines.

In summary, all randomised clinical trials of both 2vHPV and 4vHPV vaccines provide evidence of an excellent safety profile. Although both vaccines were associated with relatively high rates of ISR, particularly pain in most participants, when compared with either control vaccine or ISR reported for other routinely administered vaccines, these reactions were predominantly of short duration and resolved spontaneously. Systemic adverse events, such as headache and malaise, did not consistently occur more often in HPV vaccine recipients and the incidence of serious adverse events, including new onset chronic conditions and deaths, was not different in HPV vaccine recipients than in controls subjects. Whilst these data are robust, there are limitations inherent in clinical trials and meta-analyses. They are not powered to assess differences in specific chronic and/or autoimmune conditions, proposed as being potentially vaccine-related, and are not conducted in persons wholly representative of the population (for example by ethnicity or underlying medical conditions). As such, post-licensure studies in large populations have been important (discussed in Sect. 5.4).

4 Safety of HPV Vaccines During Pregnancy

HPV vaccines are not recommended for use in pregnant women and pregnant subjects were excluded from participating in the clinical trials of both HPV vaccines. However, some participants did become pregnant despite using contraception and undergoing pregnancy testing prior to dose administration. The relatively large number of pregnancies that occurred, the prospective nature of the studies, control groups and documentation of outcomes make these clinical data very valuable [71]. Pregnancy outcomes for vaccine trial participants were reported irrespective of the relationship between conception and dose/s, but also separately for participants who conceived within either 30 or 90 days of vaccination. For the majority of pregnancies, vaccination did not occur during the period when any theoretical risk would be most biologically plausible (close to the time of conception or during early embryogenesis). Thus, these studies are still limited by their small sample size. Following vaccine registration, pregnancy registries for both vaccines have been established [72, 73]. These registries collect spontaneously reported cases of exposure, which can assist in detection of a significant concern. However, they also have many limitations, such as limited knowledge of precise gestational age at time of exposure, selective reporting, underreporting, inability to calculate reporting rates (in vaccinated and unvaccinated individuals) and representativeness (in comparison with the overall pregnant population).

4.1 The 4vHPV Vaccine and Pregnancy

Overall, the proportions of pregnancies that resulted in an adverse outcome (spontaneous abortion, late foetal death, infant with congenital anomalies) among the 4vHPV vaccine and control recipients who became pregnant at any time during the clinical trials course (13–16 % of >12,000 participants aged 15–26 years) were similar [74]. In one study, where 70 women were vaccinated within 30 days of conception, there were 5 cases of congenital abnormalities in infants of vaccine recipients compared with none among 66 women who received aluminium-containing placebo, with a statistically significant risk difference (4.5; 95 % CI 1.1–10.1) [28]. However, those abnormalities were relatively common and unrelated: hip dysplasia, congenital ankyglossia/congenital pyloric stenosis, congenital hydronephrosis, congenital megacolon and talipes. A subsequent pooled analysis of five phase 3 clinical trials, for outcomes following vaccination within 30 days of conception, identified no additional cases among vaccine recipients but one infant with congenital anomalies born to a placebo recipient. This rendered the risk difference statistically insignificant [74]. Over all time periods, the number of infants with congenital anomalies was not statistically different between the vaccine and placebo groups (40 versus 30; 2.0 versus 1.5 %, p = 0.20) and rates were consistent with those expected. Although generally reassuring, caution in drawing firm conclusions from this pooled analysis is warranted, in part because of various methodological limitations, particularly post-hoc pooling of studies of different design [71].

Post-licensure data on pregnancy outcomes reported to the pregnancy registry for 4vHPV established by the manufacturer are published covering a 2-year period [75]. These included 517 prospectively followed pregnancies, with 451 (87.2 %) resulting in live births, including three sets of twins [72]. Of these HPV vaccine-exposed neonates 439 (96.7 %) were normal. The rates of spontaneous abortions [6.9 per 100 outcomes (95 % CI 4.8–9.6)] and major birth defects were not greater than unexposed population rates. The prevalence of major birth defects was 2.2 per 100 live-born neonates (95 % CI 1.05–4.05). The registry also received reports of two cases each of the rare conditions anencephaly and schizencephaly (one reported prospectively and one retrospectively for each condition) in infants whose mothers received the vaccine within 2–21 days after the last menstrual period [72]. The estimated population prevalence of these two conditions was 1.1 per 100,000 births and 1.5 per 100,000 births, respectively, but because of the limitations discussed above, incidence rate estimates in the vaccinated population cannot be calculated. Detailed annual reports from this registry can be obtained on request. The most recent contains data on approximately 1,500 pregnancies prospectively followed to May 2011 and does not indicate a causal relationship between 4vHPV and adverse pregnancy outcomes [75].

4.2 The 2vHPV Vaccine and Pregnancy

Similar to 4vHPV, there are no specific studies of the use of 2vHPV in pregnant women. Published studies reporting on pregnancy outcomes in 2vHPV clinical trial participants predominantly focussed on miscarriage (spontaneous abortion) rates. A combined analysis from two double-blind randomised controlled trials of 2vHPV found that among >26,000 women aged 15–25 years there were 3,599 pregnancies eligible for analysis [76]. The 2vHPV vaccine was not associated with an increase in the risk of miscarriage compared with the control hepatitis A vaccine (11.5 versus 10.1 %, respectively), and rates of miscarriage were not different from that expected. In women who conceived within 90 days after receiving a vaccine or placebo dose (n = 230), a non-significant increased rate in miscarriage in the 2vHPV vaccine recipients was observed (13.7 versus 9.2 %, p = 0.033 by permutation test) [76].

Another pooled analysis, which included participants of one of the two studies above, found that in 415 participants who became pregnant around the time of vaccination spontaneous abortion rates varied between 2vHPV and control recipients, being higher among younger 2vHPV vaccine recipients [11 % compared with those who received aluminium-containing placebo (8.3 %) or hepatitis A vaccine (5.8 %)], but lower among older 2vHPV compared with placebo recipients [42]. Rates of pregnancy resulting in premature delivery did not differ substantially. Data on congenital anomalies were not reported in these published studies [42]. A vaccine registry for 2vHPV [77] has also been established, but no published data from the registry are available.

4.3 Conclusion Regarding Use of HPV Vaccines in Pregnant Women

In summary, pregnancy adverse outcomes in both 4vHPV and 2vHPV studies appeared similar overall among vaccine compared with control recipients and were comparable to population rates. Evidence for an epidemiologic association of infant congenital anomalies or miscarriage with receipt of a dose of either vaccine at any time, including close to conception, has not been established from the available clinical trial and post-licensure data. However, as discussed above, these studies have limitations, and additional data from population-based data linkage studies of pregnancy outcomes in vaccinated women, would be valuable. For example, a study of ASO3-adjuvanted pandemic influenza vaccine in Denmark demonstrated no difference in pregnancy outcomes in vaccinated compared with unvaccinated women [78]. Immunisation guidelines recommend that pregnant women should avoid vaccination until after delivery [16, 79, 80]. Women inadvertently vaccinated whilst pregnant, or who conceive shortly after vaccination, can be reassured there is no evidence to indicate need for medical termination of pregnancy [16, 79].

5 Post-Licensure Experience

Although safety is one of the primary outcomes measured in vaccine clinical trials, the number of participants included and short follow-up periods are limitations [81] that inherently restrict the identification of rare adverse events [18]. Clinical trial participants are often homogenous with regards to age, ethnicity and health status and hence trial results are not always generalisable to the populations in which vaccines will be introduced. Post-licensure surveillance is essential to detect rare or unexpected adverse events and to monitor safety in large diverse populations under variable real-life conditions. Many post-licensure assessments of HPV vaccine safety have been performed and are discussed below [18, 82].

5.1 Passive Surveillance

Some countries have passive reporting systems for AE related to medicines and vaccines in which information is spontaneously reported by health practitioners or the public rather than systematically sought out. These include the UK Yellow Card scheme (medicines and vaccines) [83], and others specifically for vaccines, including the US Vaccine Adverse Event Reporting System (VAERS) [84], and the Canadian Adverse Events Following Immunisation Surveillance System (CAEFISS) [8587]. Reports of adverse events following immunisation (AEFI), in addition to medicines, are also collected via Vigibase, the World Health Organisation’s programme, from multiple countries [88]. However, these systems have inherent limitations including: capture of only a small proportion of total adverse events; great variation in reporting frequency, quality and completeness; lack of timely accurate data on vaccine usage (estimates of incidence rates calculated on all doses distributed rather than by age stratified data on doses administered); lack of detailed clinical data required to assess causality; and recording of all events independent of whether a causal association with the vaccine administered exists [89]. Identification of safety signals for unexpected or previously unknown events may arise [90]; however, determining causality is often difficult [91] and usually requires further evaluation using additional surveillance and epidemiological studies [92].

Published national passive surveillance data for HPV vaccination are available from Australia, the Netherlands, USA and UK [9396]. As expected, reporting rates differ (Table 1) because of differences in reporting mechanisms, case definitions and how rates are derived. For example, by the end of 2008, the US VAERS received 12,424 reports of adverse events following more than 23 million doses of the 4vHPV vaccine distributed, giving an overall reporting rate of 539 reports per million doses distributed [94]. In comparison, reporting rates for adverse events following 4vHPV in Australia were 249 per million doses distributed, but were higher (~400 per million) if limited to school girls only [97, 98]. With 2vHPV, higher reporting rates for adverse events were seen in the UK (1,045 per million doses administered) and in the Netherlands following a vaccination campaign for girls 12–16 years of age (1,160 per million doses administered); however, this may represent system differences rather than differences in AE rates per se. [95, 99]. Reporting rates cannot be calculated for data from multiple countries using WHO’s Vigibase because of the lack of data on doses distributed or administered [99].

Table 1 Reporting rates for adverse events following HPV vaccination from passive safety surveillance systems

The majority of AEFIs with the HPV vaccination reported via passive surveillance have been minor and align with expected adverse events seen in pre-licensure clinical trials. The most commonly reported AEFI included injection site reactions, headache and dizziness [9396, 100]. Only a small and expected proportion of AEFIs were categorised as ‘serious’, for example, 7 % of all reports to VAERS [101]. The majority of reported events to date are in females: in the 2 years after the 4vHPV vaccine was registered in males in the US, there were 504 male reports made to VAERS (6.5 % considered serious). The most common non-serious adverse events reported by males were similar to females and included dizziness, syncope and injection site pain [101].

A myriad of more serious adverse events following HPV vaccination have been reported to passive surveillance systems, including anaphylaxis, Guillain-Barré syndrome, transverse myelitis and thromboembolic events [94]. Comparisons of observed versus expected incidence rates of disease in a population are subject to the limitations of the passive reporting systems discussed above and the availability of data on background rates for a specific condition. Thus, enhanced passive and/or active surveillance to detect specific adverse events and/or to investigate signals derived from passive surveillance have been conducted or are ongoing. As discussed below, such studies utilise accurate data on vaccines administered and can better establish risk of HPV-vaccine attributable AE [82].

No deaths reported and published in passive surveillance systems data have been attributed as causally related to either of the HPV vaccines [93, 96, 101]. The importance of thoroughly investigating deaths that occur shortly after vaccination to determine potential alternative causes was highlighted by a much publicized case in the UK, where a teenage girl died suddenly on the same day of vaccination from a previously undetected tumour [102]. The propensity to assume causality with serious outcomes was also poignantly highlighted when a HPV demonstration project in females in India was suspended because of the occurrence of four deaths temporarily related to vaccination [103]. Although the deaths have been reportedly found to be unrelated to vaccination, loss of confidence in HPV vaccination led to the suspension of clinical studies of HPV vaccine in India [104].

5.2 Case Reports and Case Series

When a vaccine is first licensed, publications of AEFI reports from individuals or groups of individuals are common. However, such ‘case reports’ can only rarely provide strong evidence of a causal link with vaccination, typically when the AE that occurs is directly related to the vaccine (for example, injection site reactions or isolation of a neuropathic vaccine-derived poliovirus from a recently vaccinated patient with paralytic poliomyelitis). Even if such reports, together with other scientific data, present credible “mechanistic” evidence of the possibility of a vaccine causing the adverse event, the frequency at which these events occur in relationship to vaccination or in the absence of vaccination (that is, the epidemiologic evidence) will not be available from such studies [105].

Table 2 lists published case reports/case series of AEs following HPV immunisation. Of these, five describe a case or cases of local/regional reactions related to the injection site. Such reports may highlight the occurrence of vaccine delivery errors, the need for improvement in vaccine administration techniques and/or important management issues, for example, the need to avoid repeated vaccination into the same site at which lipoatrophy has occurred or to institute prompt multi-modal therapy for complex regional pain syndrome [106, 107]. However, other reports describe occurrences of serious or striking disease onset for conditions whose aetiology and/or pathogenesis is uncertain. It is well recognised that reporting and attribution of such cases to vaccination can be anticipated when a new vaccine is introduced into a population, particularly where large-scale programmes with high coverage and multiple vaccine doses are implemented [108]. However, implicating vaccines as a causal factor for such diseases, for example multiple sclerosis [109], requires more than a temporal association. Analysis of disease rates between vaccinated and unvaccinated individuals, a consistent temporal relationship, biological plausibility and lack of alternative aetiology are essential factors to be considered [105, 108]. Population-based studies and other evidence available to date, discussed more below, provide evidence that autoimmune diseases are not triggered by HPV vaccination.

Table 2 Published case series/case reports of adverse events following HPV immunisation

Table 2 also contains reference to other events found not to be biologically related to vaccination, such as the occurrence of a “mass psychogenic” response to vaccination among girls in a school-based vaccination clinic in Australia. This report highlighted that factors other than the vaccine constituents, such as the fear of painful events, can trigger adverse reactions to vaccination [20, 110]. Expert review of unusual and serious conditions that occur after vaccination, such as that conducted in specialised immunisation adverse event assessment clinics, is important [111, 112].

5.3 Enhanced Passive and Active Surveillance

Table 3 describes four post-licensure studies that used survey methods to investigate adverse reactions following HPV vaccination predominantly in teenage girls in four different settings: the Netherlands and Italy (2vHPV) and the USA and Nepal (4vHPV). The quality of these studies, particularly the response rate, varied. Overall, the rates of ISR and systemic reactions reported following vaccination were similar to that seen in randomised controlled clinical trials, with the exception of the limited data from the study in Nepal [113], where only 8 % of girls reported pain at the injection site following 4vHPV. The methods of data collection were not presented in this study and it appears likely that methodological and socio-cultural factors may have limited perception and/or reporting of pain.

Table 3 Post-licensure studies of adverse events following HPV vaccine using enhanced passive surveillance methods

The most comprehensive study was in the Netherlands, where 4,248 girls aged 13–16 years (74 % of those approached) responded [95]. Participation rates declined progressively after the first, second and third doses of vaccine were received; however, overall no unexpected or SAEs were reported (Table 3). Reporting rates for ISRs were comparable to those for the 2vHPV vaccine in clinical trials, although a higher proportion of participants subjectively reported “pronounced pain” at the injection site (24, 12 and 15 % by dose, respectively). ISRs and systemic symptoms were significantly less common after second and third doses, although this did not account for factors such as non-completion of vaccine course or the survey. Girls who reported feeling unwell in the week prior to vaccination, or who had a history of certain chronic medical conditions, had a statistically higher likelihood of reporting AEs post-vaccination. In a similar study among Italian school girls, no SAEs were reported [114]. ISRs, myalgia, fatigue and headache were the most common reported symptoms. Pain was more commonly reported after the first dose, whereas rates of other local and systemic reactions were higher after subsequent doses. AEs were less frequently reported than in published clinical trials. A smaller study from the US, with a lower response rate but utilising medical record review to also ascertain AE, found girls reported similar ISRs compared with clinical trial data [115]. This study was notable for high rates (n = 134, 15 %) of syncope and dizziness (recorded as “pre-syncope”) after vaccination.

5.4 Population-Based Epidemiologic Surveillance

Table 4 lists seven published studies of SAE following HPV vaccination in well-defined large populations. Three of these studies focussed on specific conditions: anaphylaxis [116], Guillain-Barré syndrome [117] and thrombocytopenia [118] in children following vaccination of any type. The other five publications report on observational cohort studies conducted using large healthcare databases in the USA [119122] or France [123] and examined the incidence of pre-specified new onset conditions following vaccination. The study from France (on 2vHPV vaccine) was presented in abstract format only and could not be fully scrutinised. In the US studies, potential new onset conditions were identified by various methods (including hospital/emergency department discharge diagnosis codes, pharmacy prescriptions and laboratory data) and cases had in-depth clinical review. Comparison with incidence rates in unvaccinated same-age female populations (concurrently or historically) or rates following vaccines other than 4vHPV were made. Two studies report on females aged 9–26 years between August 2006 and October 2009 in the Vaccine Safety Datalink (VSD) cohort, in which a total of 600,558 doses of 4vHPV vaccine were recorded [120, 121]. There was no statistically significant increased risk in vaccine recipients for the outcomes studied [these included GBS, stroke, appendicitis, seizures, allergic reactions, anaphylaxis, syncope and venous thromboembolism (VTE)]. The other two US studies [119, 122] were based on a cohort of almost 190,000 females aged 9–26 years who had received at least one dose of the 4vHPV vaccine. There was no cluster in the onset of 16 pre-specified new onset autoimmune diseases (NOAID) over an 180-day period in relationship to vaccination, age or dose(s) received [119]. Further investigation of an elevated incidence rate ratio for one of the 16 conditions, Hashimoto’s disease, did not suggest a true association, as many of the cases were likely pre-existing at the time of vaccination. A more extensive review of medical attendance for 265 distinctly coded conditions occurring in various plausible risk periods following vaccination indicated that only syncope (on the day of vaccination) and ‘skin infections’ (likely representing injection site reactions) occurred more commonly after vaccination than during non-exposure periods [122].

Table 4 Post-licensure population-based studies of adverse events following HPV vaccine

With the introduction of new vaccines such as HPV vaccines, where three doses are scheduled over a 6-month period, there will inevitably be the onset of new conditions, including immune-mediated diseases, associated in time with vaccination [124]. Rates of immune-mediated diseases in the female population targeted for HPV vaccination (9–26 years) are relatively high, with disease incidence varying substantially across this age group [124]. For example, the hospitalisation rate for autoimmune conditions in women 19–30 years is up to 389/100,000 [62], and in US women aged 19–28 years diagnoses of thyroiditis and multiple sclerosis are ten times more frequent than in adolescent girls [108]. Studies to determine “background rates” of diseases suggested to be potentially vaccine-attributable have been conducted in some countries, such as Denmark [125] and the US [62, 108], and are key to understanding and addressing vaccine safety concerns when they inevitably arise.

5.5 Conditions of Interest Arising from Post-Licensure Surveillance

5.5.1 Syncope

Disproportionately higher rates of syncope after HPV vaccination compared with that observed in pre-licensure RCTs were reported from post-licensure surveillance in the US [115], Australia [94], the Netherlands [126] and Italy [114]. The reporting rate for syncope (Table 1) following 4vHPV was similar in the USA and the Netherlands and in state-based but not national reports from Australia, at approximately 80–100 per million doses [127]. Although neither dizziness nor syncope was reported at increased rates in vaccine compared with control recipients in pre-licensure studies, given the high overall reporting rate of pain following vaccination, syncope in response to this painful stimulus is not unexpected under real-life conditions. It has become clear that in the adolescent vaccine target population, reports of syncope following any vaccination are common compared with other target groups [128]. In the US, syncope in persons >5 years of age following any vaccination increased from 0.28 per million vaccine doses distributed in 2004 to 0.54 per million doses distributed in 2006 after the introduction of three new vaccines targeted to adolescents: meningococcal conjugate vaccine, reduced dose diphtheria-tetanus-pertussis vaccine and HPV vaccine [129]. Analysis of VSD data found no increased risk of syncope specifically following HPV4 when compared with other vaccines given to adolescents [120]. Syncope can lead to syncopal seizures [127] and falls: about 15 % of syncopal episodes reported to VAERS resulted in a fall, of which 68 % resulted in a head injury [94]. This highlights the importance of practical measures to anticipate and/or avoid syncope and its complications, such as lying down during, and resting after, vaccination [128, 129]. In one Australian review, a majority of patients with HPV vaccine-related syncope or syncopal seizures were re-vaccinated without recurrence [127].

5.5.2 Anaphylaxis

Although considered a rare AEFI, anaphylaxis can occur because of antigen, vaccine adjuvant and/or vaccine excipients [98]. Of note, no cases of vaccine-related anaphylaxis were reported in the phase 3 studies of HPV vaccines, although episodes of immediate and severe allergic reactions (facial oedema, severe bronchospasm) occurred [18]. Reporting rates of anaphylaxis following HPV vaccination have generally been consistent in both national passive surveillance systems and population-based studies (Tables 1, 3) and within the estimated range of that for other vaccines of 1–10 cases per million doses. The exception was one early study from Australia in which higher rates of anaphylaxis (2.6 per 100,000 doses) were reported in a school-based program [97]; this could partly be explained by the different surveillance mechanism and case definitions employed in this study. Of note, the 4vHPV vaccine is produced in a yeast-based (Saccharomyces cerevisiae) system and use of this vaccine is contraindicated in persons with a history of immediate hypersensitivity to yeast. The 2vHPV vaccine pre-filled syringes contain latex within the stopper and should not be used in persons with a history of anaphylaxis to latex [130]. Overall, the evidence from post-licensure surveillance suggests that anaphylaxis and serious allergic reactions following HPV vaccination are rare and manageable, consistent with that found for other vaccines. In two retrospective cohort studies in females from Australia, only a proportion of those reported as ‘suspected’ anaphylaxis cases were classified as anaphylaxis on clinical review, and even fewer (1/19) were found to have probable hypersensitivity after skin prick testing [97, 98]. The majority of girls with a history of suspected anaphylaxis who were re-vaccinated under close medical observation had no subsequent adverse reactions [98].

5.5.3 Guillain-Barré syndrome (GBS)

A review of GBS cases reported to the US VAERS estimated a 2.5- to 10-times greater rate of GBS within 6 weeks after 4vHPV vaccination compared with that expected in the general population (6.6 events per week per 10 million subjects versus 0.65–2.57 cases per week per 10 million population) [131]. However, the increased propensity to report events that occur in a temporal relationship to vaccination and lack of a control group meant further investigation to better delineate any possible relationship was required. Subsequent population-based studies, employing extensive case finding methods for GBS, have not provided evidence of a rate that is significantly greater than that expected in the adolescent and young adult female population (Table 4) [119, 120].

5.5.4 Venous Thromboembolism (VTE)

A safety signal for VTE was identified following review of VAERS reports to June 2006 [94]. However, of the cases that had sufficient information for review, 90 % had a known risk factor for VTE, such as oral contraceptive use. The time to onset after vaccination was also highly variable [94]. Analysis of VTE in the VSD population-based cohort revealed a non-significant increased relative risk of 1.98 among females aged 9–17 years when compared with historical rates; however, all five cases had known risk factors for VTE [120]. Increased rates of VTE post HPV vaccination have not been reported in other settings, suggesting this is unlikely to be causally related to vaccination.

6 Summary of Findings

A variety of population-based post-licensure studies have assessed adverse events, including those with delayed onset, in females given HPV vaccines. Despite the inevitable publication of case reports raising the potential association between HPV vaccination and a range of severe chronic conditions of poorly defined aetiology such as multiple sclerosis, evidence from these well-conducted population-based studies has consistently not identified any new or concerning safety issues. Randomised controlled clinical trials conducted before vaccine licensure identified that ISR (particularly pain) and mild self-limited systemic symptoms (such as myalgia and headache) occur commonly after HPV vaccination and should be anticipated. One head-to-head comparison study found injection site reactions and some systemic symptoms to be more common in 2vHPV compared with 4vHPV vaccine recipients although, overall, events in most recipients were well-tolerated and self-limiting. Allergic reactions and syncope can occur following HPV vaccination but can be managed with appropriate care [132]. Data from pooled clinical studies of HPV vaccination given inadvertently during pregnancy or near the time of conception have not provided evidence for foetal harm or higher rates of miscarriage; however, advice regarding avoiding vaccination during pregnancy is warranted, given the limitations of such studies.

The findings of this review are consistent with the expert assessment of a number of peak advisory bodies [13, 96, 133]. For example, the Global Advisory Committee of the World Health Organisation found in 2007 and 2008 that current evidence on the safety of HPV vaccines was reassuring [132, 134]. Independent systematic reviews of HPV vaccine safety have also drawn this conclusion [18]. In addition, a recent report by the Institute of Medicine (IOM) summarising evidence for causal relationships between certain adverse events and eight different vaccines, including HPV vaccines, identified acceptable levels of evidence to support a causal relationship between HPV vaccination and anaphylaxis and reported acceptance of a causal relationship with any vaccine injection and syncope [105]. Consistent with the findings of this review, the IOM stated that neither the mechanistic or epidemiologic evidence supported an association of HPV vaccination with other outcomes, such as various new onset chronic diseases.

7 Conclusions and Future Considerations

This overview provides an update of the continually expanding body of evidence regarding the safety profile of the two currently licensed HPV vaccines, 2vHPV (Cervarix®) and 4vHPV (Gardasil®). Both have been well characterised in extensive clinical trials and a range of post-licensure studies, some of the most comprehensive employed for any vaccine. Overwhelming, the evidence supports an excellent safety profile, not unlike the majority of other inactivated vaccines assessed in similar populations.

Ongoing initiatives to assess vaccine safety have been summarised elsewhere [82, 135]. Despite the reassuring evidence to date, safety studies in developing countries that adopt HPV vaccine use, and in specific populations not extensively assessed to date, such as males, are awaited. For example, studies in immunocompromised persons, an important target group, are limited and should occur. In parallel, research to better elucidate background rates of serious diseases and events that can be coincidentally associated in time with vaccination are essential to inform decision-making around vaccine safety issues as they arise [124] and to avoid a loss of confidence in these potentially life-saving vaccines.