Is It Safe to Implant Prosthesis When Frozen Section Reports Positive During Joint Arthroplasty? A Retrospective Cohort Study

ABSTRACT

Background

Periprosthetic joint infection (PJI) remains a devastating complication of joint arthroplasty. Although intraoperative frozen section histology (> 5 neutrophils/HPF) is incorporated into the 2018 MSIS diagnostic criteria, its prognostic value for predicting postoperative PJI remains controversial.

Methods

Primary or revision THA/TKA procedures in our hospital between January 1, 2015 and December 31, 2021 with intraoperative frozen section histologic results were eligible for inclusion. Association between intraoperative frozen section histology and postoperative PJI was investigated based on the medical records, and a similar analysis of C‐reactive protein (CRP) and erythrocyte sedimentation rate (ESR) was also performed.

Results

A total of 14,969 knee/hip procedures were identified. After case selection, 797 cases comprising 529 (66.4%) primary and 268 (33.6%) revision were included. According to our data, frozen section positivity showed no PJI association, both in primary (OR, 1.206 [95% CI, 0.133–10.914], P > 0.999) and revision procedures (OR, 1.287 [95% CI, 0.331–5.007], p = 0.717). Conversely, elevated CRP (OR, 15.196 [95% CI, 1.916–120.523], p = 0.001) and ESR (OR, 5.794 [95% CI, 1.4979–22.4129], p = 0.008) predicted PJI exclusively in revisions.

Conclusions

Therefore, isolated positive frozen section results should not be the sole basis for surgical strategy. Elevated CRP/ESR levels during revision surgery warrant heightened vigilance and consideration of extended anti‐infective treatment.

Level of Evidence

III.

This study demonstrated that it is improper to overinterpret isolated frozen section positivity. It may be safe to implant prosthesis in patients who had positive frozen sections but without other signs of infection. Surgeons opting for staged revisions solely based on neutrophil counts risk unnecessary morbidity, particularly in patients with inflammatory comorbidities.

1. Introduction

Periprosthetic joint infection (PJI) is one of the most severe complications following total knee arthroplasty (TKA) or total hip arthroplasty (THA) [1, 2]. The reported incidence rate of PJI following TKA or THA is 0.7%–2.4% [1, 3, 4]. Although uncommon, PJI may be potentially fatal and lead to disability, pain, and poor quality of life [5, 6]. Currently, the increasing number of TKAs or THAs is accompanied by a number of PJIs [7]. PJI usually requires a revision procedure to control infection and replace the implants to restore joint function, which increases the economic burden on the patients and public healthcare system [8, 9].

Preoperative identification of patients with high PJI risk facilitates surgical targeting and additional prevention to lower the incidence rate of PJI [3]. Several factors including male sex, smoking, increasing body mass index (BMI), steroid use, previous joint surgery, and comorbidities, such as diabetes, rheumatoid arthritis, and depression are risk factors for PJI [1, 10]. Although still controversial, serologic markers such as erythrocyte sedimentation rate (ESR) and C‐reactive protein (CRP) levels were reported widely to be associated with increased risk for PJI [11, 12, 13].

Performing prosthesis implantation in an actively infected joint increases the risk for PJI. Intraoperative frozen section histology was reported to be effective for detecting active infection in TKA/THA revision [14, 15, 16]. Intraoperative findings of positive histology scored three points in Musculoskeletal Infection Society (MSIS) 2018 criteria for PJI [17]. Thus, surgeons may opt for a staged arthroplasty because of positive intraoperative histology. Despite the MSIS endorsement, mounting evidence suggests neutrophil infiltration in frozen sections may reflect noninfectious synovitis [18, 19, 20, 21]. The predictive value of intraoperative frozen section histology in PJI remains contentious. Ahmadi et al. reported a high specificity and low sensitivity of intraoperative histology for predicting infection in revision elbow arthroplasty [22]. Bori et al. demonstrated the increased possibility of PJI in positive intraoperative histology [14]. This discrepancy may stem from anatomic site variability or different neutrophil count thresholds. Although the 2018 MSIS criteria incorporated frozen section analysis to enhance PJI diagnostic accuracy, our data challenge its prognostic utility: neither primary nor revision arthroplasties exhibited significant associations between positive histology and subsequent PJI. This aligns with Straub et al.'s conclusion that frozen sections during two‐stage reimplantation poorly predict reinfection [23].

Therefore, intraoperative frozen section analysis may not be reliable. This study hypothesizes that: (1) Frozen section positivity poorly predicts postoperative PJI due to inflammatory confounders; (2) CRP/ESR demonstrate procedure‐specific value in PJI prediction.

2. Materials and Methods

2.1. Patients and Data

All data presented in this retrospective study was based on the medical records of patients hospitalized in our hospital between January 1, 2015 and December 31, 2021. All patients undergoing THA or TKA were eligible for inclusion. Both primary and revision arthroplasties were included. Frozen section analysis was not routinely performed in primary TKA/THA. It was selectively utilized in primary cases at high risk for PJI based on preoperative assessment and routinely in all revision procedures. Demographic data (age, gender), intraoperative histological tissue frozen section results, ESR and CRP values before operation, medical history, and follow‐up records were extracted.

2.2. Case Selection

Inclusion criteria:

1. THA/TKA performed for noninfectious indications (including primary and revision procedures);

2. Intraoperative histology results based on the number of neutrophils per high‐power field (HPF):

3. For staged procedures, only the definitive prosthesis implantation was included.

Exclusion criteria:

1. Aseptic cases with < 2 years of follow‐up;

2. Surgery performed for neoplastic disease;

3. Cement spacer placement due to positive histology;

4. Administration of additional antibiotics.

All included cases had no clinical signs of local infection preoperatively. All included patients underwent prosthesis implantation as planned. Cases where the surgical plan was altered due to frozen section findings (e.g., abandonment of prosthesis implantation or insertion of a cement spacer) were excluded from analysis, as our aim was to evaluate the outcome of ignoring an isolated positive frozen section. Diagnosis of postoperative PJI was made according to the 2018 MSIS criteria. Infection cases were patients who experienced subsequent PJI. Aseptic cases were those who had no PJI during the follow‐up phase.

For infection cases, the follow‐up phase ended on the date of diagnosis for PJI. The last follow‐up time in our medical records was the endpoint of the follow‐up phase for aseptic cases. The routine perioperative antibiotic regimen in our department is intravenous antibiotics once before surgery and twice daily within 48 h after surgery, and antibiotic species include cefuroxime, cefazolin, and clindamycin. Application of antibiotics other than these antibiotics and medication time beyond 48 h after surgery were all defined as additional antibiotics. Additional anti‐infective treatment would potentially decrease the risk of PJI, which introduces bias. All cases of one‐stage revision for PJI were excluded because additional antibiotic therapy was applied in those cases. However, if the definitive prosthesis implantation in staged revision for PJI did not exceed routine perioperative antibiotic therapy, those cases were included in this study.

2.3. Blood Tests

Tests of ESR and CRP level were routinely performed before joint arthroplasty in our hospital. The test results were collected. ESR < 30 mm/h and CRP < 10 mg/L were regarded as negative test results accordingly [1, 10].

2.4. Intraoperative Histology

In this study, intraoperative histology refers to rapid histologic evaluation of frozen sections obtained from intraoperative samples of synovial tissue in primary procedures or periprosthetic tissue in revisions. Histologic result of frozen sections was collected: a positive intraoperative histologic result was based on > 5 neutrophils per HPF; while a negative result was < 5 neutrophils per HPF accordingly [17, 24]. Only the maximum number was obtained for patients with multiple frozen sections.

2.5. Statistical Analysis

Postoperative infection rates of positive and negative intraoperative histology groups were calculated and compared using the Fisher's exact test. p Value > 0.05 was regarded as statistically similar. Odds ratio (OR) was also calculated and presented with a 95% confidence interval. We also performed similar statistical analysis of CRP and ESR values. Furthermore, the association between positive intraoperative frozen section and elevated CRP and ESR was analyzed using the same method.

In order to analyze the association between positive intraoperative histology and autoimmune or noninfectious comorbidities, positive intraoperative histology rates of two common comorbidities (gout and rheumatoid arthritis) were calculated and compared with the other included cases.

3. Results

A total of 14,969 knee/hip procedures (14,321 [95.67%] primary and 648 [4.33%] revision) were performed in our hospital between January 1, 2015 and December 31, 2021. After case selection (Figure 1), 797 cases (529 [66.4%] primary and 268 [33.6%] revision) comprising 331 (41.5%) knees and 466 (58.5%) hips were included (Figures 2 and 3). The total number included 445 (55.8%) female and 352 (44.2%) male patients. The average age at surgery was 63.3 years with a standard deviation of 11.9. In summary, 16 of 797 (2.0%) suffered from postoperative PJI. The PJI rate associated with primary procedure was 0.95% (five of 529), and the PJI rate associated with revision procedure was 4.1% (11 of 268). The mean follow‐up time of aseptic cases was 30.97 months, ranging from 24 to 83 months postsurgery. The mean time to PJI in infection cases was 8.00 months, ranging from 1 to 32 months postsurgery. The shorter mean time to PJI in infection cases reflects the endpoint of observation upon diagnosis, whereas all aseptic cases were monitored for a minimum of 24 months, a period sufficient to observe the vast majority of PJI events. Details of each included case are presented in Table S1.

FIGURE 1.

Flow chart of this study.

FIGURE 2.

A 69‐year‐old male (PT‐562) underwent revision knee arthroplasty (A). Preoperative laboratory tests showed elevated inflammatory markers: CRP 82.4 mg/L and ESR 46 mm/h. Two intraoperative frozen sections were obtained, both of which were negative for acute inflammation (< 5 polymorphonuclear neutrophils per high‐power field). The patient received a prosthesis (B and C) and developed PJI at 1 year postrevision (D and E). The PJI was successfully managed with debridement, antibiotics, and implant retention (DAIR) treatment.

FIGURE 3.

An 89‐year‐old male (PT‐462) with a history of gout underwent primary total knee arthroplasty (A and B). Preoperative laboratory tests showed elevated inflammatory markers: CRP 86.3 mg/L and ESR 71 mm/h. Intraoperative frozen section was strongly positive, demonstrating > 30 polymorphonuclear neutrophils per high‐power field. Despite these concerning findings, the patient did not develop any clinical or microbiological evidence of PJI during follow‐up (C and D).

3.1. Association Between Positive Intraoperative Frozen Section and Incidence of Postoperative PJI

Among the 152 cases of positive frozen section (> 5 neutrophils/HPF), four (2.63%) were diagnosed with postoperative PJI. While 645 cases reported negative results (< 5 neutrophils/HPF), 12 (1.85%) of them had postoperative PJI, which was statistically similar to the PJI rate of the positive group (OR, 1.426 [95%CI, 0.454–4.483], p = 0.523). Baseline data showed no statistical difference in gender and age between the frozen positive and negative groups. Subgroup analysis data was presented in Tables 1 and 2: for primary procedures, PJI rate of positive frozen section group (one of 91, 1.09%) was not significantly higher than negative group (four of 438, 0.91%). OR, 1.206 [95%CI, 0.133–10.914], P > 0.999; similar result was obtained in revision arthroplasties (OR, 1.287 [95%CI, 0.331–5.007], p = 0.717).

TABLE 1.

Association between intraoperative frozen section and postoperative PJI.

	Frozen section positive	Frozen section negative	p
Age (years, mean ± SD)
Primary procedure	63.1 ± 11.7	61.5 ± 12.5	0.262
Revision procedure	67.9 ± 8.5	65.9 ± 10.6	0.178
Gender (female/male)
Primary procedure	50/41	233/215	0.609
Revision procedure	35/26	127/80	0.577
Periprosthetic joint infection rate
Primary procedure	1.09% (1/91)	0.91% (4/438)	> 0.999
Revision procedure	4.92% (3/61)	3.86% (8/207)	0.717
Total	2.63% (4/152)	1.85% (12/645)	0.523

TABLE 2.

Odds ratio (OR) for PJI.

	OR with 95% CI
	Primary cohort	Revision cohort	Total cohort
Positive frozen section	1.206 [0.133–10.914]	1.287 [0.331–5.007]	1.426 [0.454–4.483]
CRP ≥ 10 mg/L	2.173 [0.360–13.115]	15.196 [1.916–120.523]*	6.377 [1.802–22.559]*
ESR ≥ 30 mm/h	2.573 [0.426–15.532]	5.794 [1.4979–22.4129]*	4.071 [1.400–11.836]*

^* p < 0.05.

3.2. Association Between Positive CRP Level and Incidence of Postoperative PJI

A total of 329 cases tested positive for elevated CRP levels, while 13 (3.95%) of them had postoperative PJI. Negative CRP test results were obtained with 468 case, whereas three (0.64%) of them had postoperative PJI, which was statistically lower than the PJI rate of the positive group (OR, 6.377 [95%CI, 1.802–22.559], p = 0.001). Subgroup analysis data was presented in Tables 2 and 3: for primary procedures, PJI rate of positive CRP levels (three of 217, 1.38%) was not significantly higher than negative group (two of 312, 0.64%. OR, 2.173 [95%CI, 0.360–13.115], p = 0.405); but PJI rate of revision cases with elevated CRP level (10 of 112, 8.92%) was significantly higher than the others (one of 156, 0.64%, OR, 15.196 [95%CI, 1.916–120.523], p = 0.001).

TABLE 3.

Association between CRP/ESR level and postoperative PJI.

Cohort	Biomarker and threshold	Periprosthetic joint infection rate	p
Primary procedure
	CRP ≥ 10 mg/L	1.38% (n = 217)	0.405
	CRP < 10 mg/L	0.64% (n = 312)
	ESR ≥ 30 mm/h	1.53% (n = 196)	0.365
	ESR < 30 mm/h	0.60% (n = 333)
Revision procedure
	CRP ≥ 10 mg/L	8.92% (n = 112)	0.001
	CRP < 10 mg/L	0.64% (n = 156)
	ESR ≥ 30 mm/h	8.99% (n = 89)	0.008
	ESR < 30 mm/h	1.68% (n = 179)
Total
	CRP ≥ 10 mg/L	3.95% (n = 329)	0.001
	CRP < 10 mg/L	0.64% (n = 468)
	ESR ≥ 30 mm/h	3.86% (n = 285)	0.008
	ESR < 30 mm/h	0.98% (n = 512)

3.3. Association Between Positive ESR and Incidence of Postoperative PJI

A total of 285 cases tested positive for elevated ESR, while 11 (3.86%) of them had postoperative PJI. Negative ESR test results were obtained for 512 cases, whereas five (0.98%) of them had postoperative PJI, which was statistically lower than in the positive group (OR, 4.071 [95%CI, 1.400–11.836], p = 0.008). Subgroup analysis data was presented in Tables 2 and 3: for primary procedures, PJI rate of positive ESR levels (three of 196, 1.53%) was statistically similar with negative group (two of 333, 0.60%. OR, 2.573 [95%CI, 0.426–15.532], p = 0.365); but PJI rate of revision cases with elevated ESR level (8 of 89, 8.99%) was significantly higher than the others (three of 179, 1.68%, OR, 5.794 [95%CI, 1.4979–22.4129], p = 0.008).

3.4. Association Between Positive Intraoperative Histology and Autoimmune or Noninfectious Comorbidities

Among 797 selected cases, 152 (overall rate: 19.07%) reported positive frozen section. Fifty‐one cases had a history of rheumatoid arthritis, and 21 of them (41.18%) reported positive frozen section. In 20 with gout, eight cases (40.00%) reported positive frozen section. Among patients with inflammatory comorbidities (RA or gout, n = 71), the frozen section positive rate was significantly higher than the others (n = 726) (40.85% vs. 16.94%; p < 0.0001).

To quantify this association, we calculated the odds of a positive frozen section. The OR for having a positive frozen section in patients with RA or gout, compared to those without, was 3.42 (95% CI: 2.05–5.69, p < 0.0001). This indicates that the presence of these inflammatory comorbidities was associated with an increased odds of intraoperative frozen section positivity.

3.5. Association Between Positive Intraoperative Frozen Section and Elevated CRP/ESR

Among patients with positive frozen sections, elevated CRP (≥ 10 mg/L) and ESR (≥ 30 mm/h) were observed in 71.71% (109/152) and 65.79% (100/152) of cases, significantly higher than in the frozen section‐negative group (34.11%, p < 0.0001 and 28.68%, p < 0.0001). Subgroup analysis data (Table 4) showed that this significant association between positive intraoperative frozen section and elevated CRP/ESR existed in both revision and primary procedures.

TABLE 4.

Association between positive intraoperative frozen section and elevated CRP/ESR.

Cohort	Biomarker and Threshold	Frozen section positive	Frozen section negative	p
Primary procedure		n = 91	n = 438
	CRP ≥ 10 mg/L, n (%)	73 (80.22%)	144 (32.87%)	< 0.0001
	ESR ≥ 30 mm/h, n (%)	67 (73.62%)	129 (29.45%)	< 0.0001
Revision procedure		n = 61	n = 207
	CRP ≥ 10 mg/L, n (%)	36 (58.06%)	76 (36.71%)	0.0016
	ESR ≥ 30 mm/h, n (%)	33 (54.10%)	56 (27.05%)	0.0001

3.6. Association Between Inflammatory Arthritis and Elevated CRP/ESR

The preoperative CRP and ESR levels of cases with inflammatory arthritis (RA and gout) in primary surgeries were analyzed. Patients with inflammatory arthritis had significantly higher median CRP and ESR levels compared to those with the others (CRP: 21.5 vs. 5.8 mg/L, p < 0.001; ESR: 42 vs. 18 mm/h, p < 0.001). This baseline elevation explains the reduced specificity of these markers for predicting PJI in primary surgeries.

4. Discussion

Intraoperative frozen sections are used widely in our hospital to detect infection in revision TKA/THAs as suggested by published studies [25, 26, 27]. However, it is not a routine test in a primary procedure, and usually be used in patients at increases risk for PJI. Primary procedures at increased risk for PJI have the following characteristics: (1) BMI > 30, (2) active smoker, (3) ASA ≥ 3, (4) immunosuppression (i.e., diabetes mellitus, rheumatoid arthritis, hepatitis C virus, human immunodeficiency virus, etc.), and (5) Staphylococcus aureus colonization [1, 28]. Typically, the surgeon selects the most suspicious tissue for intraoperative histology to exclude the possibility of local infection. As a result, although more than 1000 primary TKA/THAs were performed in our hospital every year, the number of primary procedures meeting our criteria is limited. In order to enlarge the sample size and study the role of frozen section histology in primary and revision procedures, both procedures were included.

4.1. Pathobiological Rationale for False‐Positive Histology

Neutrophil infiltration, usually regarded as an infection‐specific marker [29], may also involve autoimmune and noninfectious inflammatory disorders [19, 20, 30]. Our findings substantiate this hypothesis: cases with preexisting diagnoses of these comorbidities exhibited significantly higher positive frozen section rates. It should be noted that the missed diagnosis of rheumatic disease may aggravate this confounding factor due to the lack of systematic screening and patient awareness, which suggests noninfectious inflammation frequently interferes with histologic infection criteria. Even in aseptic prosthetic joint loosening, particulate debris from implant wear may trigger localized neutrophil recruitment and inflammation [31], which would possibly cause false‐positive histology in revision procedures.

The high false‐positive rate of frozen sections can be associated with similar molecular pathways between infection and sterile inflammation. Crystals in gout, immune complexes in rheumatoid arthritis, and implant wear debris can activate the inflammasome [31, 32], leading to pyroptosis of synovial cells. This programmed cell release releases IL‐1β and other potent neutrophil chemoattractants, recruiting and satisfying histologic infection criteria without any pathogen [33, 34]. Conversely, some pathogens may confound this response. For instance, S. aureus can suppress pyroptosis in bone cells to establish persistent, low‐grade infection that may not trigger a dense neutrophil infiltrate [35]. This duality may explain the poor predictive value of isolated frozen section positivity and underscores the necessity of multimodal assessment.

4.2. Inflammatory Markers: Divergent Predictive Value

CRP and ESR exhibited procedure‐dependent prognostic significance. Elevated CRP was associated with a 15‐fold increased PJI risk in revisions (p = 0.001) but lacked association in primary arthroplasties (p = 0.405). Similarly, high ESR predicted PJI in revisions (OR 5.79, p = 0.008) but not primary cases. This discrepancy may be related to baseline inflammation: Chronic synovitis in osteoarthritis drives baseline CRP/ESR elevation through IL‐6 and TNF‐α secretion from hypertrophic synovium before primary TKA/THA [36, 37]. Synovectomy during primary arthroplasty removes this cytokine source. Consequently, CRP/ESR elevation in revisions more specifically reflects infection‐triggered inflammation.

4.3. Interplay Between Local Neutrophil Infiltration and Systemic Inflammation

Our data reveal a significant association between positive intraoperative frozen section and elevated systemic inflammatory markers (CRP/ESR). While such elevations typically signify active inflammation, our data demonstrated that frozen section positivity showed no independent association with PJI risk, suggesting local neutrophils may instead herald sterile inflammatory processes. This phenomenon aligns with emerging evidence of neutrophil plasticity [21]. Beyond infection, neutrophils mediate tissue remodeling in osteoarthritis [38], perpetuate autoimmune synovitis in rheumatoid arthritis [39], and respond to crystalline deposits in gout [20]. Critically, surgeons must differentiate “inflammatory noise” from true infection—a task achievable only through multimodal assessment.

4.4. Clinical Implications

Based on our findings, we propose the following approach: For primary procedures: Isolated positive frozen section without other clinical or laboratory signs of infection should not preclude prosthesis implantation; for revision procedures: Preoperative CRP ≥ 10 mg/L or ESR ≥ 30 mm/h should be taken seriously, even if frozen section is negative. In such cases, we recommend extended postoperative intravenous antibiotics (e.g., 5–7 days) based on local antimicrobial resistance patterns, and close follow‐up is also necessary; Frozen section should be interpreted as one component of a multimodal assessment that includes preoperative markers, intraoperative findings (purulence, sinus tract), and microbiological culture.

4.5. Strengths and Limitations

4.5.1. Strengths

(1) It represents one of the largest single‐center cohorts (n = 797) examining the association between intraoperative frozen section histology and postoperative PJI, with detailed subgroup analyses for primary and revision arthroplasties; (2) We evaluated the confounding effect of inflammatory comorbidities (rheumatoid arthritis and gout) on frozen section results, which have been rarely quantified in prior studies.

4.5.2. Limitations

(1) Although excluding tumor cases reduced confounding, it limits generalizability to oncologic arthroplasty, which has a significantly higher infection risk [40, 41]; (2) Exclusion of cement spacer placement in staged procedures and cases with additional antibiotics strengthened internal validity but limits generalizability to complex infections; (3) The retrospective design limits causal inference; (4) Surgeon‐dependent technique variability may confound outcomes, though all of them are from one center; (5) The low PJI incidence requires validation in larger cohorts to rule out Type II errors.

4.6. Conclusions

In this retrospective cohort study, we found that intraoperative frozen section positivity (> 5 neutrophils/HPF) was not significantly associated with an increased risk of subsequent PJI, either in primary or revision procedures. In contrast, elevated preoperative CRP and ESR demonstrated strong predictive value in revision arthroplasty where they conferred a 15‐fold and 5.8‐fold increased PJI risk.

Author Contributions

Lingyun Wu: methodology, project administration, resources. Chiyuan Ma: conceptualization, investigation, writing – review and editing. Jiaxuan Zou: visualization, methodology, project administration. Jieli Zhou: writing – original draft, investigation. Pengfei Hu: investigation. Pingan Lin: validation, formal analysis, supervision, data curation. Haobo Wu: resources, writing – review and editing, methodology, funding acquisition, software. Lidong Wu: writing – review and editing.

Funding

This study was supported by grants from the National Natural Science Foundation of China (NO. 82001458).

Ethics Statement

This study was approved by the Ethics Committee of the Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China (2023–1207). All participants gave informed consent prior to participation. This study was conducted in compliance with the Declaration of Helsinki and all applicable ethics guidelines.

Consent

All participants gave informed consent prior to participation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Details of each included case.

Data Availability Statement

The data supporting the findings of this study are available within the article and its Table S1.