Patient demographics

The ‘pre-2007’ group contained 45 implants retrieved from 21 male and 24 female patients, with a median age of 61.3 years (interquartile range (IQR) 57.1 to 65.5); the ‘2007 onwards’ group contained 47 implants retrieved from 19 male and 28 female patients, with a median age of 61.8 years (IQR 58.5 to 67.8).

The reasons for revision were unexplained pain confirmed as ARMD post-revision (n = 88), infection (n = 1), femoral loosening (n = 1), malposition (n = 1), and recurrent dislocations (n = 1).

Measurement of bearing surface material loss

The volume of bearing surface material loss from the retrieved Pinnacle implants was measured using a Zeiss Prismo (Carl Zeiss Ltd, Rugby, UK) coordinate measuring machine (CMM). Utilizing previously published protocols,¹⁰ a 2 mm ruby stylus followed a number of polar scan lines on each surface, collecting up to 30 000 data points. An iterative least-squares fitting method was used to analyze the data, and wear maps were produced through the comparison of the unworn geometry and the bearing surface. This provided a means of illustrating the position of worn regions and identifying the presence of edge wear.

Measurement of head taper material loss

A rectangular representation of each taper surface was produced using a Talyrond 365 roundness measuring machine (Taylor Hobson Ltd, Leicester, UK), from which topographical features were identified and volumetric material loss quantified. This reconstruction was the result of combining 180 vertical traces recorded by a 5 μm diamond stylus, which translated along the taper axis of each femoral head. Previously published protocols were used to formulate this method.¹¹

Statistical analysis

First, statistical analysis was used to compare the patient, implant, and surgical characteristics of both groups. These included gender, age, time to revision, stem design, inclination, horizontal offset, vertical offset, and edge wear. The categorical variables were compared using Fisher’s exact test, while all continuous characteristics were compared using the unpaired t-test. The second stage of the analysis compared the bearing wear rate and taper wear rate of the two groups. Allowing for potentially confounding factors, the following comparisons were performed using linear regression: 1) unadjusted, not considering possible confounding factors; and 2) adjusted, adjusting for patient/hip characteristics.

To limit the number of variables in the analysis, only the characteristics found to be different between groups (p < 0.1) were adjusted for. Both study outcomes were analyzed on the log scale. A small constant was added before the transformation so that patients with zero wear rates could be included in the analysis. As a result of the log transformation, the wear rate of the ‘2007 onwards’ group relative to the wear rate of the ‘pre-2007’ group was obtained as a ratio. The corresponding confidence intervals were also reported, along with p-values indicating the significance of the results.

We confirm that all investigations were conducted in conformity with ethical principles of research, that informed consent for participation in the study was obtained, and that institutional approval of the human protocol for this investigation was obtained.

Comparison of surgical, implant, and patient factors

The two groups in this study were comparable, in regard to age (p = 0.77), gender (p = 0.67), horizontal femoral offset (p = 0.23), and edge wear (p = 0.08) (Table I). All other compared characteristics were somewhat different, as highlighted in Table I. These included the time to revision, paired stem design, inclination, and vertical offset.

In regard to the individual whole blood measures of Co and Cr, the difference between the two groups was not statistically significant; however, the Co/Cr ratio was found to be significantly greater in the ‘pre-2007’ group (Table II).

Comparison of wear rates

The median bearing wear rate of the ‘pre-2007’ and ‘2007 onwards’ groups was 2.3 mm³/year and 3.8 mm³/year, respectively (p = 0.39). The median taper wear rate of the pre-2007 implants was 0.81 mm³/year, while the ‘2007 onwards’ group had a median rate of 0.19 mm³/year (p = 0.005) (Table III). These comparisons are presented in Figures 1 and 2.

Fig. 1

Graph showing the comparison of the bearing wear rate of the ‘pre-2007’ and ‘2007 onwards’ groups.

Fig. 2

Graph showing the comparison of the taper wear rate of the ‘pre-2007’ and ‘2007 onwards’ groups.

There was no difference in the bearing wear rate between the groups in either the unadjusted regression analysis or after adjusting for possible confounding factors (time to revision, stem design, edge wear, cup inclination, and vertical offset; Table IV). However, taper wear rate was significantly lower in hips implanted from 2007 onwards, when no other factors were taken into account (unadjusted analysis). On average, material was lost from these tapers at only around half the rate recorded from the ‘pre-2007’ group. Nevertheless, after adjusting for potentially confounding factors (as with the bearing wear rate analysis), the difference in taper wear rate between the groups was not statistically significant, as seen in Table IV.

Implant time to revision

In this study, Pinnacles implanted from 2007 onwards were revised significantly earlier than in the ‘pre-2007’ hips (Fig. 3). As both groups were found to have comparable wear rates, this trend could be attributed to the increased surveillance of MoM hips in the later years, rather than implant performance. Increased vigilance after 2007 would likely be associated with the coinciding high coverage and attention received by MoM implants, for example, due to the failures that led to the recall of ASRs (DePuy Synthes) in 2010.¹² Additionally, as advised by the Medicines and Healthcare products Regulatory Agency (MHRA), from 2012 onwards, patients with these MoM hip arthroplasties would have been monitored annually, allowing for earlier detection of ARMD with improved imaging modalities including metal artefact reduction sequence (MARS)-MRI.¹³

Fig. 3

Box plots representing the time to revision for both groups of implants.

A reduced threshold for revision surgery could be another contributor to this trend, resulting in more recent implants being removed from patients who were possibly not previously considered for such revision treatment. In 2010, the MHRA advised surgeons that whole blood metal ion levels greater than 7 ppb were to be considered in order to identify patients who required additional follow-up.¹⁴ This may have influenced their consideration of revision surgery even when dealing with asymptomatic patients.¹⁵

The authors of a recent study using Pinnacle revision rates from the NJR⁶ may have been misled by the same trend in the absence of wear measurements. The addition of these data can provide a better insight into implant performance; however, it can only be obtained through retrieval analysis. This evidence raises doubts regarding the use of ‘revision rate’ and ‘time to revision’ as reliable indicators of implant performance, especially in the analysis of hips with large ranges of implantation dates. Figure 4 clearly demonstrates an annual reduction in the mean time to revision as the year of Pinnacle implantation becomes more recent. As the trend is also evident between 2003 and 2007, a change in the performance of the implants from 2007 onwards could not be reliably identified through the analysis of this parameter.

Fig. 4

Bar chart plotting the year of primary implantation against the mean time to revision.

Whole blood metal ions

Cobalt and chromium ions released from Pinnacle implants during wear can be identified in the blood. The presence of these particles can cause necrotic and inflammatory changes in the periprosthetic tissue,¹⁶ including ARMD,¹⁵ prompting symptoms that often justify implant revision. Patients in this study demonstrated levels consistent with typical revision cases while no significant difference was seen between the two groups. The significantly greater Co/Cr ratio found in the blood of patients from the ‘pre-2007’ group is consistent with their greater taper wear rates in the unadjusted analysis, as corrosion in this area is known to result in elevated Co blood levels compared with Cr.¹⁷ A previous study into Pinnacle implants suggested that taper wear rate accelerates over time,¹⁸ which may explain the lower rate of material lost from this area in examples implanted since 2007. Increasing vertical femoral offset has also previously been shown to be a risk factor for greater corrosion.¹⁸

A limitation of this study was the use of Pinnacle implantation dates as a surrogate for their date of manufacture, since the discrepancy between these events can vary depending on the length of time the implants remain unused. Without an exact date for the speculated change in Pinnacle manufacture, the date of 1 January 2007 was used to separate the two groups, which may have resulted in some affected hips being included in the pre-2007 group. This classification method was, nevertheless, consistent with previous investigations into the same trend. As with all retrieval studies, we do not know the as-manufactured dimensional state of the components prior to implantation and how this would have impacted the individual clearances of each implant.

Another possible limitation is the number of variables that differed between the two groups, which could have had a confounding effect. However, statistical measures were taken in order to account for the characteristics that were significantly different between the groups, and all analysis was performed by a medical statistician. While this is one of the largest retrieval studies to have investigated this topic, we acknowledge that all studies are at some risk of Type II errors; this may be mitigated in future studies that involve larger sample sizes.

In conclusion, this study found no significant difference in the bearing wear rate of Pinnacle hips implanted before 2007 and from 2007 onwards. Therefore, the hypothesized decline in their performance after 2006 could not be supported. The time to revision was found to decrease consistently as the year of primary implantation increased. We speculate that this is due to stricter MoM implant surveillance in later years. Future research into the performance of these hips should be wary of this trend, especially in the absence of retrieval data.