The value of inflammatory markers to diagnose and monitor diabetic foot osteomyelitis

Abstract

In this study, we assessed the effectiveness of inflammatory markers to diagnose and monitor the treatment of osteomyelitis in the diabetic foot. We evaluated 35 consecutive patients admitted to our hospital with infected foot ulcers. Patients were divided in two groups based on the results of bone culture and histopathology: osteomyelitis and no osteomyelitis. The erythrocyte sedimentation rate (ESR), C‐reactive protein (CRP), procalcitonin (PCT), interleukin‐6 (IL‐6), interleukin‐8 (IL‐8), tumor necrosis factor alpha (TNFα), monocyte chemotactic protein‐1 (MCP‐1) and macrophage inflammatory protein‐1 alpha (MIP1α) were measured at baseline after 3 and 6 weeks of standard therapy. PCT levels in the osteomyelitis group were significantly higher at baseline than in the group with no osteomyelitis (P = 0·049). There were no significant differences between the two groups in the levels of the other markers. CRP, ESR, PCT and IL‐6 levels significantly declined in the group with osteomyelitis after starting therapy, while MCP‐1 increased (P = 0·002). TNFα and MIP1α levels were below range in 80 out of 97 samples and therefore not reported. Our results suggest that PCT might be useful to distinguish osteomyelitis in infected foot ulcers. CRP, ESR, PCT and IL‐6 are valuable when monitoring the effect of therapy.

Introduction

As part of the worldwide epidemic of diabetes mellitus, the prevalence of lower extremity complications is rising. Patients with diabetes now have a 25% lifetime risk of developing a foot ulcer 1. Foot infections are one of the most common reasons for hospitalisation and lower extremity amputations in patients with diabetes 2, 3. Up to 68% of the persons admitted to the hospital with a diabetic foot infection have diabetic foot osteomyelitis (DFO) 4, 5.

Osteomyelitis is difficult to accurately diagnose and to determine when treatment is successful so antibiotics can be discontinued 6, 7. The ramifications of inaccurate diagnosis of DFO can be dire. An incorrect diagnosis of DFO when only soft tissue infection is present (i.e. without infected bone) leads to prolonged antibiotic therapy, unnecessary surgery and amputation 8. Prolonged antibiotic therapy can contribute to bacterial resistance, acute kidney injury, catheter‐related infections and gastrointestinal complications, such as Clostridium difficile infection 9, 10. On the other hand, failure to recognise the presence of DFO in case bone infection is actually present may lead to poor wound healing, progression of infection and increased risk of developing systemic infections 11.

Diagnosing DFO and monitoring the resolution of infection remains a challenge with very little evidence that biomarkers or imaging techniques can accurately assess the presence of bone infection or monitor the effectiveness of therapy. Bone culture and histology are considered the gold standard to diagnose DFO in most guidelines 12, but it is not routinely performed in clinical practice. The erythrocyte sedimentation rate (ESR) has been suggested to be the best available laboratory test to diagnose and monitor DFO 13. Only a few small studies have evaluated the value of C‐reactive protein (CRP), procalcitonin (PCT) and cytokines to diagnose and monitor osteomyelitis in a diabetic foot 14, 15, 16, 17. Unfortunately, the quality of the available studies is low, patient populations are small and the heterogeneous use of reference tests to confirm DFO makes it difficult to compare data 18.

In this study, we aimed to assess the value of inflammatory markers in clinical practice to distinguish osteomyelitis from soft tissue infection and to determine the role of these markers in monitoring therapy.

Methods

Study design

We conducted a prospective cohort study comparing biomarkers in patients admitted to our hospital with diabetic foot infections over a 4‐month period. Patients were followed‐up for 6 weeks and received standard of care, including surgical treatment when needed. During the baseline visit, we collected patient demographic information and medical history and performed an extensive wound, neurological and vascular examination. We evaluated the status of peripheral arterial disease through the examination of the ankle brachial index using a portable Doppler machine (Koven Technology Inc., St. Louis, MO, USA); values <0·9 were considered abnormal. We measured segmental skin perfusion pressure and pulse volume recordings using a Sensilase Pad‐IQ (Väsamed, Eden Prairie, MN, USA) system 19. We also assessed the peripheral neurological status of both feet using monofilament sensory and vibration threshold perception tests.

We consecutively included patients who were 21 years or older and had a moderate or severe infected ulcer, based on the Infectious Diseases Society of America (IDSA) classification, with a suspicion of osteomyelitis 6. Suspicion of osteomyelitis was based on infection severity, clinical presentation, radiographic changes, magnetic resonance imaging (MRI) findings, the probe to bone test or deep infection near the bone or joint. We excluded patients with other infectious diseases, previously diagnosed but still active bone infection in the study foot, immunosuppressive therapy, organ and/or haematological malignancies and end stage renal disease requiring dialysis. The baseline diagnosis of osteomyelitis was confirmed by positive histopathological examination and/or culture of bone. To determine if osteomyelitis was still present after surgery, we examined the surgical margin of resected bone (if surgery was performed). Typically, we treated patients empirically with a combination of vancomycin and piperacillin/tazobactam. We switched from empirical antibiotic therapy to a targeted therapy based on the sensitivity results of the cultures. Written informed consent and ethics committee approval was obtained before the start of the study.

Biological parameters

We drew blood at baseline after 3 and 6 weeks of therapy to determine the levels of the inflammatory markers: erythrocyte sedimentation rate (ESR), C‐reactive protein (CRP), procalcitonin (PCT), interleukin‐6 (IL‐6), interleukin‐8 (IL‐8), tumor necrosis factor alpha (TNFα), monocyte chemotactic protein‐1 (MCP‐1) and macrophage inflammatory protein‐1 alpha (MIP1α). CRP and ESR were analysed by the hospital's biochemistry laboratory. We centrifuged the blood after 30 minutes of collection for 12 minutes at G force of 1811.16 (radius 18 cm, 3000 rpm). 0·1 ml of venous plasma was stored as a batch at −80°C for later analysis to minimise variance. The PCT concentration in serum was measured with an automatic kryptor device using a BRAHMS procalcitonin kit (BRAHMS Diagnostica, Berlin, Germany). The functional sensitivity is reported to be 0·04 ng/ml with an intra assay variation lower than 5% 20. A custom‐designed human inflammatory cytokine immunoassay using Luminex^® Technology (Thermo Fisher Scientific Inc., Waltham, MA, USA) was used to analyse cytokine levels. Laboratory technicians were kept unaware of the clinical data.

Statistical analysis

Analysis was performed using the SAS 9.4 statistical package (SAS Institute Inc., Cary, NC, USA). We assessed differences between the two groups using parametrical or non‐parametrical methods according to the specific indications. Differences between the laboratory levels of the two groups at baseline were measured using the Wilcoxon rank‐sum test in median. Data were presented as mean ± standard deviation. In addition, analysis of variance for repeated measurements was performed to test the timing effect of the study parameters during the follow‐up period. The same analysis was used to examine differences during follow‐up between patients with and without DFO. The Greenhouse–Geisser adjustment was used when the sphericity assumptions were not fulfilled. P values < 0·05 were considered statistically significant.

Results

Table 1 presents an overview of the patient characteristics. We enrolled 36 patients at baseline; one patient left the hospital against medical advice before we could perform the bone biopsy. Twenty‐four patients had a positive diagnosis of osteomyelitis (DFO group) and 11 patients had no evidence of osteomyelitis in the bone that was examined (NDFO group). Patient characteristics were not different between patients with and without DFO, except for the male/female ratio (P = 0·03) and a history of previous foot ulcers (P = 0·007). The diagnosis of the 24 patients in the DFO group was based on results of percutaneous biopsies in seven patients and bone biopsies during surgery in 17 patients (Table 2). The seven patients that received a percutaneous biopsy were treated medically without any surgery. From the 17 patients that had surgery, nine had clean surgical margins after initial irrigation and debridement and eight had ‘dirty’ margins, defined as continued bacterial growth.

Table 1.

Characteristics and laboratory data of the patients from each group^*

	DFO group	NDFO group	P‐value
	n = 24 (%)	n = 11 (%)
Sex, male (%)	16 (67)	11 (100)	0·03
Age (years)	50·6 ± 11·7	51·1 ± 8·1	0·11
BMI, >30	30·3 ± 5·0 (50)	28·4 ± 5·7 (36)	0·7
Tobacco
Previous use	13 (54)	6 (54·5)	0·93
Current use	3 (12·5)	2 (18)	0·66
Diabetes mellitus type 2	21 (87·5)	10 (91)	0·77
ABI; ABI < 0·9 (%)	1·02 ± 0·23 (30)	1·01 ± 0·19 (20)	0·71
SPP (mmHg)
Great toe	58·3 ± 36·9	78·73 ± 36·3	0·16
Plantar medial forefoot	69·3 ± 25·1	79·45 ± 21·9	0·26
Plantar lateral forefoot	76·3 ± 26·6	77·18 ± 32·2	0·93
Dorsal foot	73·1 ± 39·5	72·3 ± 22·1	0·95
VPT (Hz)	58·51 ± 26·5	41·94 ± 17·0	0·08
Previous DFU	18 (75)	3 (27)	0·007
Temperature at baseline (°C)	36·4 ± 0·8	36·6 ± 0·5	0·24
Depth index wound (mm)	13·0 ± 11·6	7·7 ± 5·7	0·29
Positive PTB	14 (58)	5 (45)	0·48
Antibiotics at admission (weeks)	8 (33)	1 (9)	0·22
Antibiotics duration during study (weeks)	6·29 ± 2·5	3·45 ± 2·1	0·48
WBC (109/l)	8·23 ± 3·62	7·55 ± 3·12	0·58
HbA1c (%)	9·8 ± 2·7	10·7 ± 0·5	0·30

ABI, ankle brachial index; BMI,body mass index, defined as (weight in kg)/(length in meters)²; DFO, patients with diabetic foot osteomyelitis; HbA1c, glycated hemoglobin; NDFO, patients with no diabetic foot osteomyelitis; PTB, probe to bone test; SPP, skin perfusion pressure; VPT, vibration perception threshold; WBC, white blood cell count.

^*All data were collected during admission at the baseline visit unless otherwise specified.

Table 2.

Confirmation of DFO diagnosis

Percutaneous biopsy (n = 7)
Positive culture	2
Positive culture and positive pathology	5
Surgical margin (n = 17)
Positive culture	10
Positive culture and positive pathology	6
Positive pathology	1

DFO, diabetic foot osteomyelitis.

We collected serum samples of 30 and 32 patients at week 3 and week 6, respectively. TNFα and MIP1α levels were below range in 80 of 97 samples and are therefore not reported. The PCT levels in the DFO group were significantly higher at baseline than in the NDFO group (P = 0·049). Mean baseline CRP levels were also higher in the DFO group compared to the NDFO group (10·08 mg/dl versus 5·44 mg/dl; P = 0·054, Table 3). The other markers were not significantly different between patients with and without osteomyelitis at baseline.

Table 3.

Inflammatory markers at baseline and during follow‐up

	Baseline (n = 35)	3 weeks (n = 30)	6 weeks (n = 32)	P *	P †
CRP (mg/dl)
DFO group, mean ± SD	10·08 ± 8·62	0·46 ± 0·34	0·9 ± 1·02	0·0002	0·021
NDFO group, mean ± SD	5·44 ± 7·88	1·23 ± 1·58	0·92 ± 0·94	0·096
ESR (mm/hours)
DFO group, mean ± SD	78·33 ± 35·93	47·48 ± 33·18	45·23 ± 28·83	<0·0001	0·017
NDFO group, mean ± SD	58·9 ± 40·25	61·38 ± 44·31	55·5 ± 40·83	0·375
PCT (ng/ml)
DFO group, mean ± SD	0·26 ± 0·45	0·06 ± 0·06	0·06 ± 0·06	0·048	0·179
NDFO group, mean ± SD	0·07 ± 0·07	0·06 ± 0·04	0·05 ± 0·03	0·292
IL‐6 (pg/ml)
DFO group, mean ± SD	14·54 ± 12·98	6·23 ± 9·36	4·35 ± 5·21	0·004	0·755
NDFO group, mean ± SD	20·91 ± 21·27	5·98 ± 7·02	8·13 ± 9·5	0·099
IL‐8 (pg/ml)
DFO group, mean ± SD	10·15 ± 4·64	52·57 ± 201·78	15·78 ± 29·18	0·347	0·526
NDFO group, mean ± SD	9·16 ± 4·42	8·53 ± 4·71	9·34 ± 3·68	0·525
MCP‐1 (pg/ml)
DFO group, mean ± SD	45·89 ± 27·19	52·39 ± 27·71	63·40 ± 26·22	0·002	0·092
NDFO group, mean ± SD	50·28 ± 22·49	41·10 ± 26·04	78·48 ± 70·63	0·078

CRP, C‐reactive protein; DFO group, patients with diabetic foot osteomyelitis; ESR, erythrocyte sedimentation rate; IL‐6, interleukin‐6; IL‐8, interleukin 8; NDFO group, patients with no diabetic foot osteomyelitis; PCT, procalcitonin; MCP‐1 Monocyte chemotactic protein‐1.

^*Values indicate the result of analysis of variance for repeated measurements within each group, P value for the effect of time.

^†Values indicate the result of analysis of variance for repeated measurements between the two groups, P value for biopsy results; time × group interaction.

CRP (P = 0·0002), ESR (P < 0·0001), PCT (P = 0·048) and IL‐6 (P = 0·004) of the DFO group decreased significantly between the baseline and follow‐up (Table 3), while MCP‐1 increased significantly (P = 0·002). At week 3, one of the patients with osteomyelitis showed an extremely high IL‐8 (933 pg/ml) compared with the other patients (8·55 ± 3·29 pg/ml). When we removed this outlier from the analyses, no significant difference was found between the IL‐8 levels between the two groups at week 3. Compared to the NDFO group, only CRP and ESR significantly decreased, P = 0·021 and P = 0·017, respectively.

Discussion

The objective of this study was to determine if inflammatory markers can be used to diagnose or monitor the treatment of DFO. The results of this study suggest that PCT is the best blood test to distinguish DFO from NDFO (P = 0·049) when patients are admitted to the hospital with a moderate to severe infected diabetic foot ulcer. The performance of CRP to diagnose osteomyelitis was comparable to the diagnostic performance of PCT in this study population, with a P‐value of 0·054. In an earlier pilot study by Mutluoglu et al. 21, PCT did not show statistically significant differences (P = 0·63) between 13 DFO patients and 11 NDFO patients. An important caveat to this observation might be that Mutluoglu and colleagues used MRI to diagnose DFO, so there may have been false positive diagnoses.

Other studies support the diagnostic accuracy of PCT in diabetic foot infections. Uzun et al. 15 studied the usefulness of inflammatory markers, including PCT, in detecting bacterial infection in patients with diabetic foot ulcers. The area under the receiver operating characteristic curve for clinical infection identification was the greatest for PCT (0·859; P < 0·001). Another pilot study by Jeandrot et al. 16 reported the diagnostic value of the combination of PCT with CRP (area under the curve 0·947) in the identification of infection in diabetic foot ulcers. Unfortunately, both studies did not confirm their diagnosis of diabetic foot infections with biopsies, and deep soft tissue infections were grouped together with the bone infections.

In a cohort study, Fleischer et al. 14 identified higher levels of CRP (P = 0·006) and ESR (P = 0·008) at baseline in 34 DFO patients compared to 20 patients with cellulitis. Fleischer included mildly infected ulcers (IDSA classification) rather than more severe infections that might be ‘diagnostically confused’ with osteomyelitis. Ertugrul et al. 22 prospectively evaluated 24 DFO patients and found higher levels of ESR (P < 0·001) and CRP (P = 0·001) in patients with DFO compared to patients without bone infection. In that study, 43% of the patients had a more superficial ulcer (Wagner 0–2 23) compared to our patient population. Rabjohn et al. 24 had similar pre‐operative levels of ESR in their osteomyelitis group (76·2 ± 35·7 mm/hours) and in their control group (59·2 ± 24·7 mm/hours) as our study population and found a significant difference between DFO and NDFO patients (P = 0·022). However, comparison with that population is difficult because patients without diabetes were also included in the analyses.

The inflammatory cytokines we evaluated do not seem to be useful in differentiating bone and soft tissue infections. The mean IL‐6 level at baseline of patients with DFO was lower (although not significant) in our study compared with patients with soft tissue infections. In a study by Weigelt et al. 25, 170 patients with diabetic foot ulcers had higher levels of IL‐6 (12·4 pg/ml), IL‐8 (11·4 pg/ml), MCP‐1 (291·0 pg/ml) and MIP‐1α (82·6 pg/ml) compared with 140 patients without foot ulcers. Both IL‐6 and MIP‐1α were significantly higher in patients with foot ulcers (P < 0·0001 and P = 0·008). After adjusting for infection severity, patients with ulcers penetrating to bone (University of Texas (UT) grade 3 26) had higher levels of IL‐6 (P = 0·003) compared with patients with superficial ulcers (UT grade 1 26). Additional adjustment for the presence of osteomyelitis did not alter the results.

While four of the six tested markers (CRP, ESR, PCT, IL‐6) significantly decreased in the DFO group during therapy, they did not change significantly in the NDFO group. When we compare these results to a similar study designed by Michail et al. 13, the differentiating value of the markers during follow‐up appears to be similar. In the study of Altay et al. 17, initial levels of both IL‐6 (122 ± 210 pg/ml) and IL‐8 (754·9 ± 1557 pg/ml) were very high compared with our population and similar patient groups in other studies 27. Altay et al. detected a significant decrease in the level of IL‐6 after 14 days of treatment in 24 patients with clinically diagnosed osteomyelitis (129 ± 164 pg/ml pretreatment versus 74·2 ± 202 pg/ml post‐treatment, P = 0·003), but the IL‐8 level did not change. These results are similar to our results in the DFO group. Monitoring MCP‐1 levels in patients with DFO has not been reported previously, but the increase appears to fit in the reported time frame of the influx of monocytes after the initial acute inflammation phase 28. Neutrophils are more likely to be involved in inflammation at an earlier stage in the process than monocytes, and MCP‐1 exhibits chemotactic activity for monocytes but not for neutrophils.

A limitation of our study design is the high pre‐test probability of osteomyelitis, given the moderate to severe IDSA classification, and the relatively small number of negative subjects. However, the study population represents the prevalence of osteomyelitis reported in hospitalised patients in other studies. In addition, we did not have a ‘wash out’ period with no antibiotic therapy before bone cultures were obtained. These patients were admitted to the hospital with IDSA moderate to severe infections that required antibiotics and surgery urgently as per IDSA and IWGDF treatment guidelines 6, 8. While this convention is discussed in the medical literature, there is no direct evidence that it affects culture results, and it unlikely influences histopathology results. The purpose of our study was not to assess the influence of potentially confounding factors like bacterial pathogens, surgical interventions, types of antibiotic therapy, therapy adherence and other events that might have influenced the levels of biomarkers. These variables need to be evaluated in studies with larger populations.

Early diagnosis and treatment of bone infections are essential for patients with DFO to prevent amputation and unnecessary iatrogenic complications. Currently, we do not have adequate evidence to support any inflammatory marker used to diagnose or monitor the treatment for osteomyelitis in this population. These preliminary data are promising, but larger, long‐term studies are needed to better define the role of inflammatory markers for DFO.

Author contribution

SAVvA collected clinical data and wrote the manuscript. NA collected clinical data. JLF collected data and edited the manuscript. KB was a co‐ investigator and edited the manuscript. EJGP edited the manuscript. LAL collected data and contributed to the manuscript. All authors have approved the final article. LAL is the guarantor of this work and takes responsibility for the contents of the article.