Introduction

In 1988, Modic et al. [1] characterised T1 and T2 weighted signal changes in the vertebral endplates on magnetic resonance images (MRI) most frequently referred to as Modic changes (MC) type I, II and III. MC, regardless of type, have been shown to coincide with with degenerative changes of intervertebral disc and chronic low back pain (CLBP) [2–5]. MC I refers to an edema-like signal intensity (hypointense T1 and hyperintense T2 signal), MC II refers to fat-like signal intensity (hyperintense T1 and isointense or slightly hyperintense T2), and MC III refers to a sclerosis-like signal intensity (hypointense T1 and T2. MC II and III changes are thought to represent the end stage of the degenerative process [1]. In the general population, MC I seem to have the strongest association, both to an active inflammatory process and CLBP [3–7].

Lumbar disc degeneration and herniation is regarded as a risk factor for developing MC, especially MC I [7–9]. Patients with lumbar disc herniation often have a combination of low back pain and radicular pain. They are more likely to improve from the radicular pain than the back pain after surgery, and it is difficult to predict which patients are at risk for developing CLBP after the operation. Some patients with CLBP and MC I changes are operated with total disc replacement [10, 11]. It has also been suggested that MC I is associated with hypermobility which could indicate the need for additional fusion surgery amongst these patients [12]. Previous studies with small smaples evaluating a possible association between MC and reduced recovery of back pain after lumbar microdiscectomy present conflicting results [11, 12].

The aim of this study was to investigate whether the presence of MC I on preoperative MRI scans represents a risk factor for less improvement of back pain 1 year after lumbar microdiscectomy. Such information could be used in the decision-making process prior to surgery.

Materials and methods

Patient population

This is a prospective study of 178 consecutive patients operated with unilateral microdiscectomy for a one level lumbar disc herniation. Follow up time was 1 year.

Patients who did not have preoperative MRI scans, had been operated in the lumbar region previously or were reoperated during the follow-up, were excluded (Fig. 1).

Fig. 1
figure 1

Study population

Full size image

The cohort was recruited from a clinical registry, intended for quality control and research, at the Department of Neurosurgery, University Hospital of North-Norway (UNN), between 1 January 2000 and 31 December 2003. The registry comprised all patients operated for degenerative disease in the lumbar spine during the period [13, 14].

Informed consent was obtained from all participants and The Data Inspectorate of Norway approved the registration and management of data.

Surgery

All patients were operated with microscopic technique, using Caspar self retaining retractors and with preoperative or peroperative fluoroscopy for level ascertainment.

Radiological assessments

After the follow-up period, two independent neuroradiologists (both working at another university hospital) evaluated the preoperative MRI scans using the “Nordic Modic Classification” for investigating vertebral endplate signals [7, 15]. The radiologists were blinded regarding patient history, clinical outcome and each other’s evaluations. In case of disagreement between the radiologists, the interpretation of the most experienced radiologist was used.

For each patient the MRI findings at all visible levels were characterised into four groups; no MC, MC type I, II and mixed type I/II.

MC I has been shown to have the strongest correlation to CLBP [6], we therefore split the patient population into two groups: patients with type I and/or mixed I/II changes at any level (“MC I group”) and patients without any MC I (“no MC I group”), i.e. the rest of the study population. We also made subgroup analyses between patients with or without any type of MC (type I or type II).

Data collection

The self-administered questionnaires, completed at admission for surgery (baseline) and follow up, were identical and used for outcome assessments. The baseline questionnaire contained additional questions about demographics and lifestyle issues. During the hospital stay, the surgeon recorded data concerning diagnosis and treatment. All patients were summoned for follow up visits at 12 months at an outpatient clinic. Questionnaires were distributed by mail and completed at home by the patients. An independent observer, a research nurse, responsible for all follow up visits, collected and checked all questionnaires and interviewed the patients about employment status and complications.

Outcomes

Primary outcome measure was the visual analogue scale (VAS) for back pain. Secondary outcome measures were: VAS for leg pain, Oswestry Disability Index (ODI), the EuroQol-5D (EQ-5D), self-reported benefit of the operation and employment status.

Back pain and leg pain

Leg and back pain intensity were graded by the patient in two separate 100 mm VAS. (0 = no pain).

ODI

The ODI is a functional score, based on the Oswestry Low Back Disability Questionnaire which contains ten questions on limitations of activities of daily living [16]. Each item is rated on a 0–5 point scale and transferred into a percentage score, the ODI (ref). The range of possible values is from 0 to 100 (0 = no disability).

EQ-5D

EQ-5D is a generic and preference-weighted measure of health-related quality of life (HRQL). It evaluates five dimensions: mobility, self-care, activities of daily life, pain and anxiety and/or depression. For each dimension, the patient describes three possible levels of problems (no, mild to moderate and severe). Hence, this descriptive system contains 243 (35) combinations or index values for health states [17]. We used the value set based on the main survey from the EuroQol group [18], which has been validated for this patient population [13]. Total score range is from −0.594 to 1, where 1 corresponds to perfect health, and 0 to death.

Perceived benefit of the operation

At follow up the patients were asked: “How much benefit have you had from the operation?”. The response alternatives were: “Very much”, “Much”, “Some”, “None at all” or “Uncertain”. We dichotomized the patients according to their responses into those who had a substantial benefit of the operation (“Very much” and “Much”), and those who had not (the rest) [13].

ASA grade (the American Society of Anaesthetists grading system (I–V))

ASA grade was registered for each patient by a doctor or a specialised nurse before surgery. ASA grade classifies patients according to their vulnerability, i.e. physical condition (from no to life-threatening systemic disease) [19].

Statistical analyses

We tested if within group changes in outcome from baseline to follow up were statistically significant (p < 0.05) using paired t test. Differences between groups were assessed by independent samples t test, Mann–Whitney U tests or Chi square tests depending on the distribution of the data. Central tendency is presented as mean and standard deviation (SD) when normally distributed, and as median and interquartile range (IQR) when skewed (Kolmogorov–Smirnov’s test).

Predictors for improvement in back pain were assessed using the change in VAS from baseline to 12 months of follow up as dependent variable and MC I (yes/no) as exposition variable. We adjusted for covariates obtained from baseline data (Table 1) if the covariates were judged to be clinically relevant and if baseline values differed significantly (level 0.1) between the two subgroups of the exposition variable, using a linear regression model.

Table 1 Characteristics of the cohort, and of patients with and without Modic type I changes (MC I, n = 36 and no MC I, n = 142) at baseline
Full size table

We tested inter-rater reliability for the evaluation of MC by computing Kappa values (Jacob Cohen, 1960).

The Statistical Package for the Social Sciences (SPSS for Windows, version 16.0, Chicago IL) was used in all analyses.

Results

Characteristics of the study population at baseline are shown in Table 1.

Amongst patients with MC I there were more smokers and fewer patients with high body mass index (BMI) compared to patients without MC I. No other statistically significant differences between the groups were found at baseline. There was no difference in back pain (median, IQR) between smokers (48, 41) and non-smokers (53, 44), p = 0.47. At baseline patients with any type of MC (type I or type II) had a mean (SD) age of was 44.0 (10.6) years compared to 37.3 (13.0) amongst patients without any MC, p < 0.001. For the other baseline characteristics, including smoking, there were no significant differences (data not shown).

Patients who were investigated by CT had the same amount of back pain at baseline (mean = 52.1, SD = 30.6) compared to the patients investigated with MRI (mean = 49.3, SD = 26.5), who were included in the study, p = 0.60.

Radiological assessments

The Kappa values of the inter-rater reliability of Modic change evaluations were 0.81 in both the levels of L5/S1 and L4/L5 and 0.66 in L3/L4.

104 patients (58 %) had some sort of MC and 36 (20 %) had MC I or mixed I/II MC at any level. Table 2 shows the distribution of MC seen on the preoperative high field MRIs. No MC type III were found.

Table 2 Distribution and prevalence of Modic changes at different levels at baseline (n = 178 patients)
Full size table

Outcome assessment

Both patients with and without MC I improved in all outcome measures at 12 months of follow up. Patients with MC I had less improvement of back pain, Eq-5D and ODI score, but for the ODI statistical significance was not reached (Table 3). The improvement in leg pain was equal in both groups. Five patients (2.8 %) had complications to the surgery (Table 1). The complications (n) were: superficial wound infection (1), discitis (1) and dural tear (3) (all of which were repaired preoperatively).

Table 3 Outcome measures at baseline and 12 months after surgery amongst patients with and without Modic type I changes (MC I, n = 36 and no MC I, n = 142)
Full size table

A subgroup analysis of smokers and non-smokers with MC I showed that the non-smokers had an improvement of back pain (SD), of 26.9 (25.8), whereas smokers had an improvement of only 8.1 (28.6), p = 0.009 (Fig. 2). Smokers with MC I did not have a statistically significant improvement of back pain.

Fig. 2
figure 2

Improvement of back pain (visual analogue scale) 12 months after microdiscectomy for smokers or non-smokers, with or without Modic changes type I (MC I). The whiskers represent the 95 % confidence intervals for the mean values shown for each bar

Full size image

On including the whole cohort, 115 (81.0 %) of the patients without MC I stated that they had a substantial benefit (“much” or “very much”) from the operation, compared to 24 (66.7 %) of the patients with MC I (P = 0.06). Of the non-smokers, 83 (83.8 %) reported a substantial benefit, compared to 55 (70.5 %) of the smokers (p = 0.034). Of the whole cohort, 126 (74.1 %) out of 170 patients (missing data = 8) had returned to work, and 12 (34.3 %) of patients with MC I and 32 (23.7 %) without MC I still received sickness benefit (p = 0.20). Of the smokers 25 (33.3 %) received sickness benefit, compared to 19 (20.2 %) of non-smokers (p = 0.05).

We found no difference in outcomes between patients with any type of MC and the rest of the patient population (data not shown).

Risk factors

The results of uni- and multi-variate analyses are shown in Table 4. Smoking was the only independent risk factor for less improvement of back pain. As illustrated in Fig. 2 there seems to be an interaction between smoking and MC I, probably best explained by an additive effect of these two risk factors.

Table 4 Negative predictors for improvement in back pain at 12 months after surgery
Full size table

Discussion

Our principle finding is that patients with MC I who are operated for lumbar disc herniation have less improvement of back pain 12 months after surgery, compared to those who have no or other types of MC. However, in the multivariate analyses, this negative association no longer showed statistical significance when adjusted for smoking, which remained the only independent risk factor. The radicular pain improved significantly in all patients, regardless of the presence of preoperative MC I or smoking habits. Our study confirms the results of previous studies, that the Nordic Modic Consensus Group classification has a high inter-observer reproducibility [15].

Two recent studies have evaluated the influence of Modic changes on improvement of LBP after lumbar microdiscectomy; Chin et al. [11] found a trend toward greater improvement in patients without any MC. Ohtori et al. [12] recently compared outcomes after microdiscectomy amongst patients with or without MC I, but found no statistically significant difference. These studies were probably underpowered and smoking habits were not evaluated. Another cohort study found no association between MC I and recovery of non-specific back pain after non-surgical treatment, but again smoking habits were not reported [20].

This study indicates that both smoking and MC I seem to have a negative impact on recovery of back pain after surgery. There seems to be an interaction between smoking and preoperative MC I, and they might affect recovery by different mechanisms. The study was not designed to evaluate the aetiology of MC. Our results should be interpreted with caution and regarded as hypothesis generating. The sample size might have been too small to recognise MC I as an independent risk factor for persisting back pain. Our findings should therefore be evaluated in a larger cohort.

Two studies of healthy Scandinavians reported an association between the prevalence of MC and smoking [5, 21], but in a later Danish study, smoking did not seem to predispose for developing MC [9] and the exact biological mechanism behind development of MC remains obscure. In our study neither MC I nor smoking was associated to back pain at baseline. This is not surprising, since the presence of disc herniation could overshadow the LBP due to MC I.

In contrast to other studies, we found no association between Body mass index (BMI) and MC [22]. Heavy physical workload has been linked to development of MC [21]. We had no exact data on work load, only on educational level which was evaluated in the multivariate analyses. Another weakness of our study is that 42 % of the patients operated were excluded because they only had preoperative CT scans and not MRI. Many of the patients were admitted from other hospitals in rural areas, where MRI was not readily available. We cannot rule out that this might have introduced a selection bias. However, our data concerning frequency and distribution of MC seems to be in accordance with the findings of previous studies [7], and we found no differences in back pain at baseline between patients investigated with CT and with MRI.

We have no MRI scans at follow up which could be valuable for evaluating the influence of smoking on radiological findings, e.g. transformation of MC I. This could also be addressed in further studies.

The presence of preoperative MC I did not influence the improvement of radicular pain, for which the patients primarily were operated, and most of the patients with MC I had a significant improvement of back pain. Patients with MC I would therefore be good candidates for microdiscectomy alone. When evaluating the need for additional surgery for these patients, it is important to acknowledge the interaction between smoking and MC I, and that smoking is a modifiable lifestyle factor. Several studies have shown a negative effect of smoking on outcomes after low back surgery including more complex procedures, and some studies indicate that smoking cessation may reduce CLBP and improve outcomes [23–29]. In general, patients operated with lumbar microdiscectomy could be advised to quit smoking, especially in cases of concomitant MC I changes, but the efficacy of a smoking prevention programme has yet to be evaluated.

Conclusions

Patients with preoperative MC I can expect less but still significant improvement of back pain 1 year after microdiscectomy, but not if they smoke cigarettes.