In the present study the frequency of mSpD was 12.7% and there were no differences in the clinical presentation, symptoms, morbidities to patients suffering from uSpD. However, the absence of pain in the affected region especially in the thoracic and cervical region highlight the need for an entire spine MRI to exclude non-contiguous SpD. Currently, for pyogenic SpD there is no clear recommendation for whole spine MRI to exclude non-contiguous infected segments. MRI is a non-invasive, non-ionising imaging technique.
It is well known that patients with tuberculous SpD can have non-contiguous infected segments of the spine. We also know that cervical and thoracic involvement is more frequent in tuberculous SpD then in patients with unspecific SpD (13). For this reason, it is undisputed that in patients with tuberculous SpD, the entire spinal column should be imaged by MRI, even if the patients have no pain at other levels of the spine (14). In a study by Abbara et al.(14), they could not detect mSpD in patients with pyogenic SpD and so did not recommend total spine MRI in patients with unspecific SpD. In response to this study, Siam et al. discussed the study population of Abbara, and pointed out that, the study size of 19 cases of unspecific SpD is too small and noted that it was not reported whether the patients had had a total spine MRI. As a result, Siam et al. recommend a whole spine MRI be performed for all patients with SpD (7).
Two studies have evaluated the percentage of multiple non-contiguous infected segments as 6.8% (15) and 13% (10) in patients with pyogenic SpD. Our data showed that 12.7% of pyogenic SpD cases were mSpD. Taken together, these high numbers of mSpD in all three studies support the conclusion that an MRI of the entire spine must be strongly recommended.
Similar to previous studies our patient cohort was geriatric (mean age uSpD = 70 years, mSpD = 71 years) and men were more frequently affected (16) but there was no evidence for a higher frequency of mSpD due a higher age.
Although it is well known that SpD is more common in patients with secondary diseases such as diabetes and renal insufficiency (17), it is unclear whether mSpD is associated with the severity of comorbidities. In the present study, we measured the severity of comorbidities via the CCI.
The CCI of patients with uSpD and mSpD were similar. In this light, it seems rational to examine all SpD patients for mSpD, as there does not appear to be an easy metric for distinguishing between the two groups. Although generally the number and severity of comorbidities are linked to poorer patient outcomes, our data contradict the hypothesis that there is a higher risk of mSpD associated with these factors (18, 19).
Back pain is the cardinal symptom of SpD (20). Shiban et al.(21) reported back pain in 100% of SpD patients and Valencius (22) in 72% of SpD patients. In our cohort 85% (n=57) of patients with uSpD reported pain. However, in the mSpD group only 20% patients indicated pain at all infected spinal regions.
Therefore, we recommend that local MRI to detect SpD should be extended to the entire spine. However, it must be taken into account that other non-infectious pathologies may be detected as well. Radiologists and referring clinicians need to be aware of potential SpD mimics including Modic Type I degenerative changes and other pathologies such as trauma or metastatic disease (23) (24) (25). For correct diagnosis it is necessary to know the specific patterns of each of these diseases and be able to distinguish them from infectious SpD. Not least, it also remains the task of radiology to evaluate image acquisition during the scan and to adapt the sequence protocol, including the recommendation to image the total spine when a focus of infection is detected (please note also the appendix).
An additional cost factor results merely from the increased measurement time, usually not from patient repositioning or additive contrast agent. The extended MRI time is a critical point that must be taken into account when focusing on limited MRI resources. On the other hand, the relevant gain in information must be considered, especially in light of the fact that an early detection and a possible early therapy start is crucial for the outcome in case of an SpD infection.
Only patients in the uSpD group (8.7%, ASIA A-D) presented with a neurological deficit at admission. The literature gives contradictory information in this regard. In some studies, up to 54% of cases with SpD have been associated with functional neurological deficits (21, 26). However, there are also studies with a smaller number of cases that associate SpD with extremity weakness (22%) which is more consistent with our results in this study (27). This could be explained by the fact that SpD was detected in our patients at an earlier stage than in previous reports. Since none of the patients in the mSpD group had a neurological deficit, this symptom does not appear to be indicative of mSpD. Thus, neurological deficits do not appear to indicate the involvement of multiple spinal regions in SpD.
The short time from symptom onset to clinical admission is a likely reason for the low incidence of neurological deficits in our cohort.
There is no common classification to categorize the image changes in SpD. However, Pola et al. published a simple classification based on MRI images of SpD in 2017 (6). We used the Pola-Classification to classify the patients in our study as all study participants had an MRI. In our study we found more cases with Pola-A (62%) and less cases with Pola-C (22%) compared to the published study group from Pola et al (Pola-A 33,6%, Pola-B 18,4%, Pola C-48%) (6). We believe that this is because our patients were likely diagnosed with SpD at an earlier stage than in the cited publication.
In Patients with mSpD we found the following Pola Classification distribution pattern: n= 21 (all affected segments); Pola-A n = 14; Pola-B n = 2, Pola-C n= 5. This shows that a multifocal presentation of SpD does not seem to be associated with a higher degree of severity with regards to the Pola-Classification and is not associated with more epidural abscesses. Another reason for our results could be that we only included patients with primary SpD without previous surgery on the spine. This could explain why there are fewer epidural abscesses in our data compared to Pola et al (28)
In the 2018 published study, 23% of patients included had post-surgery SpD. Furthermore, the missing neurological deficits and bony destruction seems to be responsible for a large part of the missing classification upgrade.
We performed surgery in eight cases (80%) with mSpD, on all involved spinal regions, and in seven of eight cases as a one-stage procedure. One patient with involvement of all three areas of the spine died after surgery of the lumbar spine, preventing completion of the two-session procedure that had been planned. One patient had a two-session procedure. One other patient underwent surgery only on the lumbar spine while the infected areas at thoracic spine were treated conservatively. All patients were treated with an antibiotic post-surgery. This therapeutic strategy of early surgical therapy and antibiotics has also been supported by Tsai et al. (29). This study reported better outcomes for patients with SpD who were treated with early surgery together with antibiotics compared to antibiotics alone including shorter hospital stays, improvement in kyphotic deformities and increased quality of life. For findings requiring a conservative approach to treatment, we recommend a follow-up MRI for an evaluation of the therapy response and, if necessary, a change of therapy.
Nevertheless, the current study has some limitations. First, it is a retrospective study. As such, the decision on which treatment to pursue was not random.
A follow-up to evaluate the outcome was not the focus of this study. It was concerned with the decision of the therapeutic approach at the time of imaging.
There are also many confounding factors, such as the degree of infection, the duration of treatment and the surgical procedure, which can influence the outcome results. This work focused on the practical approach to evaluating the final diagnosis.