To date, the indication for revision surgery for SSI is based on wound presentation, supported by IMs, particularly, CRP, and, in some cases, by imaging modalities. CRP was shown to have specific post-operative dynamics in the case of SSI as a prolonged plateau and a later second peak at about POD 10 that might indicate revision surgery [8]. Nevertheless, there is no reference test using CRP that allows for safe and early SSI diagnosis.

Our data show that the critical IM for early SSI detection is IL-6. Furthermore, compared to CRP, WBC count, PCT, and TNF-α, IL-6 appears to be the only IM that can indicate SSI up to POD 7. This is consistent with observations that IL-6 is immediately synthesized in response to infection, activating an acute immune response and inducing CRP production by hepatocytes [9, 20].

Meisner [21] showed that after surgical trauma, IL-6 and TNF-α are the first IMs to increase, followed by PCT and finally, CRP. Because both the non-infection and infection groups in our study had suffered surgical trauma, there was no initial difference in IL-6 in our study. Only after the onset of early infection was there was a measurable difference in the immune response. We propose that analogous to the sequence of increases in IMs after surgical trauma, the acute immune response in infection is initiated by IL-6 [9, 20]. The value of IL-6 in SSI diagnosis was also observed by Lenski et al. [22], who showed an AUC of 0.95 for IL-6 in SSI prediction. Notably, however, in their study, only 9 patients were infected out of 89, and 8 of those 9 patients had late deep infections with discitis or epidural abscess and underwent revision surgery on average at POD 49. In line with this study of Lenski et al. [22] was that of Berbari et al. [23], whose meta-analysis showed that IL-6 generally seemed to have had the best diagnostic accuracy for prosthetic joint infection compared to CRP and WBC count but who clearly stated that early and late infections were not differentiated.

Rettig et al. [10] found that high levels of IL-6 were associated with postoperative complications after major abdominal surgery that included not only SSIs but also pneumonia, urinary tract infection, and others. In their analysis, the difference in IL-6 between the group with complications and the group without complications was observed at POD 1, which might also have been associated with the surgical trauma [21]. In addition, Rettig et al. [10] observed differences in CRPPOD3,7 and in TNF-αPOD7, while WBC count did not differ between the groups. Rettig et al. [10] did not report a multiple comparison correction that might have led to more results with lower significance levels. Although the use of multiple comparison correction is controversial [24], when applied it to our results retrospectively (Holm-Šídák method), it confirmed the highly significant effect of IL-6POD7 (P < 0.000001) as the only remaining significantly different IM.

Two patients from the infection group underwent late revision surgery (Patient 9 on POD 32 and Patient 12 on POD 43). However, at that time, the primary IMs (i.e., CRP and WBC count) were not significantly elevated, with only one peak after surgery until their initial discharge (Patient 12: highest postoperative CRP at 95.5 mg/l on POD 3, and Patient 12: highest postoperative CRP at 38.2 mg/l on POD 4). In both patients, IL-6POD7 was below the cutoff value. Thus, both patients would have been predicted false-negatively using IL-6POD7. This might indicate different IL-6 dynamics between very early and early SSIs. Both patients were placed in the infection group because a clearly visible SSI (wound secretion, wound dehiscence) had developed within the usual range for early SSI of 30 days [17]. However, revision surgery for patient 9 was prolonged by about 30 h due to capacity problems. Patient 12 had initially been readmitted to another hospital, where unfortunately, conservative treatment was started with the administration of antibiotics.

Both PCT and TNF-α have value in the diagnosis and monitoring of sepsis [25]. Nie et al. [26] found better predictive values for PCT than for CRP in patients with acute traumatic spinal cord injury. Aouifi et al. [15] showed that PCT may be more reliable than CRP as an IM for SSI diagnosis after cardiac surgery. Although we observed a significant difference between the two groups in PCTPOD7, we could not show a relevant predictive value for SSI prediction. Little is known about the predictive value of TNF-α for SSIs. We found no difference between the two groups in TNF-α, in line with the finding of Bottner et al. [27] that TNF-α was not relevant in prosthetic joint infections.

By including all IMs from all PODs, we were able to increase the AUC to 0.98, which resulted in an almost certain SSI diagnosis. Notably, though, this approach does not seem to be a realistic option in practice due to the costs involved.

Regarding the patient characteristics and comorbidities, the number of instrumented segments, the region of surgery, and the presence of hypertension significantly differed between the groups, while current risk factors, such as diabetes mellitus, obesity, use of steroids, drainage time, and operative time, were insignificant [28], possibly because the cohort was too small.

Our observed SSI rate of 12.7% is slightly higher than the range reported in the current literature, which indicates the risk of SSIs after dorsal spondylodesis to be approximately 0.7%-11.9% [29]. Given that our study was conducted at a maximum care spine center (Level 1) of the German Spine Society (DWG®) within a university medical center, it is plausible that our patient cohort exhibited higher levels of illness, more challenging surgeries, and age compared to those seen in non-maximum care centers. This is evident from the distribution of ASA and BMI scores, where a greater proportion of patients in both groups exhibited higher ASA classifications and BMIs above 25 kg/m2.

In conclusion, in this study, compared to CRP, PCT, and TNF-α, IL-6 had the highest value for early SSI diagnosis. Based on the results in this cohort, an SSI would be diagnosed if the IL-6POD7 value is 26.0 pg/mL or higher. Because most cases of superficial wound infections can be managed without surgery by administering antibiotics[30], our results indicate a false-positive rate of 6.8% and correct antibiotic administration in 61% of the patients starting on POD 7. Early initiation of antibiotics could significantly reduce the number of revisions required. Nevertheless, in light of the present results, the future role of low-cost CRP may need to be reconsidered.

This study had limitations. First, the IMs were measured before the surgery and on PODs 1, 2, 3, 5, and 7. The measurement of the IMs on all days might have influenced the results. Second, this was a single-center prospective diagnostic study with a relatively small cohort of patients. Although the sample size was calculated and the cohort was large enough, a multicenter study with more patients would have strengthened the results. Third, although no differences between non-infection and infection groups were observed in preoperative values of age, CRP, and IL-6, all these parameters were higher in the infection groups compared to the non-infection group, particularly when considering the preoperative p value of IL-6 (p = 0.051), indicating at least a trend between groups. While we endeavor to account for all comorbidities to minimize bias in the development of an SSI, the use of corticosteroids or history of splenectomy was not accounted for, and nutrition was assessed solely by BMI, without specific malnutrition scores. Thus, an impaired immune status due to immunosuppressive status, malnutrition, or aging may have contributed to perioperative inflammation. Fourth, although statistical analysis did not reveal differences between groups regarding the underlying indication of surgery, the mix of both degenerative and traumatic pathologies could have potentially influenced the outcomes.