Preoperative synovial tissue biopsy can be useful in determining the diagnosis. A number of techniques have been published for obtaining these tissue samples. These should be done in an aseptic manner in the operating room, preferably in an anaesthetised patient. These range from percutaneously using a Tru-Cut needle (Tru-Cut; Cardinal Health, Galway, UK) or arthroscopic biopsy forceps during a formal arthroscopy [4, 5]. It should be noted that if taking the tissue samples blind and percutaneously, caution should be exercised as there is a risk of damaging the arthroplasty components to the extent that one group has recommended this blind technique should be avoided in the well-fixed arthroplasty. In these cases, a formal arthroscopy may be more appropriate with a saline-filled joint at the expense of diluting the microbiology counts in the samples taken. However, these tissue samples can still be investigated for signs of inflammation in addition to culture and microbiology. These techniques have been shown to have a higher sensitivity and specificity compared to aspiration of the synovial fluid alone and C-reactive protein (CRP), particularly in the context of late PJI [5]. A further advantage of preoperative tissue sampling is in the analysis of antibiotic sensitivity of the organisms and therefore being able to tailor perioperative and postoperative antibiotic therapy to the individual patient and microorganism.

During revision arthroplasty surgery, an intraoperative frozen section histology sample can provide some useful information to the surgeon. Gram staining techniques from this tissue can be carried out. This test however has a very low-sensitivity [6, 7], poor negative predictive value, and its results did not alter the treatment of patients undergoing revision arthroplasty due to infection. It does however have high specificity, although it is recognised that the gram-positive samples are often found in those patients where the diagnosis of PJI is not in question. Frozen section analysis for a rapid intraoperative diagnosis of PJI, on the other hand, has better diagnostic accuracy than gram staining.

Histological samples are usually assessed by the pathologist on the following day after the specimen is fixed in formalin. However, occasionally the results are required more rapidly and thus a frozen section may be performed. The pathologist has to arrive at a correct decision in a time-sensitive manner. Historically, these techniques began with De Riemer in 1818, who used frozen sections in a diagnostic manner [8]. The technique continued to evolve from the 1950s with a cumbersome long process to the use of the modern-day cryomicrotome (cryostat) [9]. The cryostat is a refrigerated box containing a rotatory microtome. The tissue is frozen with an aerosol spray and is left in the cryostat for sectioning. The tissues are then stained, typically with haematoxylin and eosin (H&E). This tissue is then analysed under a light-powered microscope. The whole process with a report from the pathologist occurs usually within 20 min [10].

Although the AAOS and the Musculoskeletal Infection Society have produced guidelines which include histology in the diagnostic criteria, many still believe that intraoperative frozen section histology is of little value.

A recent meta-analysis by Tsaras et al. [11] involving 26 studies with more than 3000 patients confirmed that intraoperative frozen sections of periprosthetic tissues performed well in predicting a diagnosis of culture-positive PJI, despite having different criteria for the histological diagnosis of infection [variable number of polymorphonuclear neutrophil (PMN) per high-power field (HPF)]. This meta-analysis illustrated that the threshold of five PMNs had a diagnostic odds ratio of 52.6 compared to threshold of ten PMNs with an odds ratio of 69.8. The authors of this meta-analysis have emphasised the importance of frozen section histology in PJI, but at the same time, note further research is required to define the threshold number of PMNs per HPFs that is most predictive of PJI. They showed that the presence of acute inflammation provided a high positive likelihood ratio of 12. The absence of acute inflammation had a negative likelihood ratio of 0.23. This meta-analysis supports that frozen section analysis is a helpful investigation for supporting a diagnosis of PJI in patients with an intermediate pretest probability of PJI. For the patient with a low pretest probability of PJI, a negative frozen section result may be sufficient to exclude the diagnosis. In summary, they conclude that intraoperative frozen sections of periprosthetic tissues performed well for diagnosing PJI but only moderate accuracy for ruling out the diagnosis.

One group from the Cleveland Clinic has carried out a study supporting the use of both frozen section and the MSIS criteria at the time of second stage revision arthroplasty [12]. Although the MSIS criteria were originally developed to aid diagnosis of PJI before revision, the Cleveland group has tried to confirm the validity of using frozen sections and the MSIS criteria to diagnose ongoing infection and thus predicting failure of second stage reimplantation. The group accepts there are limitations in their study, but note that frozen section analysis at second stage revision had a high diagnostic specificity although a low sensitivity. In 2017, the same group has further assessed the value of intraoperative histology and has concluded that frozen section yields a high specificity, positive predictive value and negative predictive value and accuracy, with moderate sensitivity [13].

At the time of revision surgery, apart from the frozen section samples (a diagnostic service which may or may not be available), formal histology can and should be sent at the time of revision to aid in the diagnosis of PJI. The Cleveland Clinic group as above has shown that the discrepancy between frozen section and permanent sections is low and they are both useful adjuncts.

Histological analysis in providing evidence of infection will demonstrate acute inflammation, namely, the presence of neutrophil infiltration at the site of infection (Table 12.2). This can be seen on both frozen and paraffin-fixed samples.

Contamination during the primary procedure and secondary haematogenous invasion into the periprosthetic tissue are the most likely reasons to develop a deep-seated PJI [14]. Acute infections are usually caused by a virulent organism and show a typical histological picture of acute inflammation. In contrast, bacterial contamination with a low-virulence organism may have minimal symptoms. These low-virulence organisms are typically caused by small colony variants of staphylococcus [15]. These low-virulence organisms may contribute to the false negative results associated with frozen section analysis [16].

Based on the histomorphological criteria, four types of periprosthetic membrane were defined by Morawietz et al. in 2006 [17]. This classification has been validated [18]. This classification was based on the detection of foreign body particles, granulation tissue and PMNs (Table 12.2).

It is generally accepted that types 1 and 4 are aseptic loosening and types 2 and 3 are septic loosening.

This classification was expanded upon by including criteria for particle identification. Type 4 has been modified to include fibrous type [19]. Further revision of the consensus classification includes the following types:

Tissue samples should be obtained using sharp dissection rather than electrocautery due to the associated thermal damage [21, 22]. It should be taken before the administration of antibiotics intraoperatively.

The histological specimens should be obtained from the interface membrane, pseudocapsule and other periprosthetic tissue. In both infected and aseptic loosening, a small layer of tissue forms between the bone and the prosthesis known as the periprosthetic interface membrane. There is associated bony lysis around the implant due to micro-motion of the implant and cellular enzymolytic processes. The pseudocapsule is the tissue surrounding the effective joint space. Feldman originally described taking two tissue samples from inflamed pseudocapsule and the interface membrane [3]. However, it is the periprosthetic interface membrane that provides the most useful diagnostic information for the pathologist, yielding a sensitivity of 85% compared to pseudocapsule 42% with similar specificities [23–25]. If the surgeon is taking histology preoperatively, the more readily accessible tissue would be the joint pseudocapsule/neosynovium, which has been described above. A minimum of three different sites of biopsy are recommended in the literature [24] and confirmed more recently in the context of frozen sections [26]. The tissue samples can then be sectioned for the microscopic examination.