Upper tibial MRI vascular marks lost in early knee osteoarthritis

Osteoarthritis (OA) of the knee causes significant economic costs to society and morbidity affecting much of the ageing population [1]. Diagnosis is traditionally by clinical and radiological examination. Only in the later stages of the disease do X-rays reveal loss of joint space and deformity.

The etiology of primary OA is usually thought to be multifactorial with age-related wear and tear, overload, trauma, genetic factors, and systemic disease being recognised as major contributors. Vascular disease has not usually been considered a significant factor [2].

To outline the arterial anatomy of the bone, Trueta used barium sulphate emulsion injection and X-rays [3]. This approach was improved by Brookes who cleared the calcified tissue with acid [4]. The mapping of bony venous drainage proved to be more challenging. Retrograde venous injection of the bone is difficult because there are usually several superficial veins draining a limb contrasting with one or two supplying arteries, venous valves prevent retrograde flow, and venous shunts may operate. Only a small percentage of limb blood flow is through the bone. As a result, previous researchers have often injected the arteries and looked at the sagittal or coronal plane anatomy. They then assumed that there was a reciprocal venous pattern such as is demonstrated in Grey’s Anatomy [5]. Crock described a subchondral plexus of vessels in vertebral bodies using Indian ink and decalcification but little interest has otherwise been directed at subchondral blood vessel patterns in weight bearing joints [6, 7].

Denham realised that forces of several times body weight were transmitted across joints during ordinary activity [8]. It is possible that weight bearing might raise subchondral intraosseous pressure [9, 10]. A raised intraosseous pressure has itself been associated with pain and osteoarthritis [11‐14]. It is likely that if the subchondral region is subject to high forces or pressures during activity there could be anatomical or vascular modifications to cope [15].

We observed axial subchondral radiating marks in the proximal tibia on (proton density spectral presaturation with attenuated inversion recovery) PD_SPAIR water bright or fat suppressed axial MRI scans of the knee as in Fig. 1. A literature search failed to provide any previous description. We found that the marks were present in axial scans of several joints but for the purpose of the present work could best be studied in axial MRI slices of the relatively flat upper tibia. In this study, we were interested in the general incidence and distribution of the marks and in their association with arthritis in patients referred to our knee clinic. We studied the distribution of these marks in terms of their distance below the tibial surface and the pattern of distribution within quadrants of the axial plane. We then explored their association with early osteoarthritis of the knee assessed by Kellgren-Lawrence (K-L) grade on contemporary plain X-rays.

We have identified previously undescribed radial vascular markings on MRI scans of the upper tibia. The white lines are also visible on T2 or water bright MRI scans but he marks are best seen in axial T2 or PD_SPAIR scans. Radiological opinion was that these marks are vascular, and probably venous. Reasons given include that they are longitudinal channels, water enhancing, occasionally in tributary form, increase in size as they reach the periphery, penetrate the cortex, and appear to link with periosteal vessels. There is a slow flow rate in these vascular marks during MRI scanning in the recumbent patient. Arteries are generally narrower, and the higher flow rate makes their contents less visible. In the popliteal artery in Figs. 1 and 6, the contents of the artery are dark whereas the superficial vein contents are bright. A literature search failed to find any relevant papers that described subchondral vessels or veins or that addresses the loss of subchondral vessels associated with osteoarthritis [17‐19].

We chose the Kellgren-Lawrence method to grade osteoarthritis of the knee on plain AP X-rays because of its simplicity and our access to the corresponding plain X-rays. The majority of patients had osteoarthritis rankings in the lower K-L grades from 0 to 2. In our practice, few patients with severe osteoarthritis would be sent for an MRI scan. We acknowledge that the K-L grading system is a relatively insensitive tool for separating early grades of OA. Many patients with grade 0 or 1 osteoarthritis on plain radiographs have well-established cartilage disease on arthroscopy which is not yet apparent on X-rays [20].

The medial side is the more frequent site for primary knee OA and patients presenting in our clinic generally had more early medial osteoarthritis. The K-L grading system cannot easily differentiate between early grades of OA. The loss of vascular marks on MRI may be a means of detecting early OA. In this study of a group with varying degrees of early osteoarthritis, there were fewer vascular marks on the medial side generally and this was associated with more advanced K-L scores. We note that females usually have smaller joints than males. We acknowledge that this may be the reason for them having fewer vascular marks. However, we found that females had fewer marks and higher mean K-L scores for osteoarthritis both medially and laterally.

MRI image quality depends on many variables [21]. The MRI scanning protocol was that used routinely in our department on a 3-Tesla machine. For analysis, we used a standard PC screen such as those found on any ward based PACS system to view the images. We acknowledge that up to 75% of the data is lost in those images compared with the specialised screens available in an X-ray department.

There are other limitations to our study. The study is cross-sectional. Although we were able to establish an inverse correlation between the presence of vascular markings in the proximal tibia and the presence and severity of osteoarthritis, we cannot comment on cause and effect. We do not know if OA worsened, improved, or remained the same as this is a cross-sectional study rather than a longitudinal study. We are unable to say whether the increasing K-L grade is a cause of or is an effect of vascular mark changes or vice versa. The study is based on an unselected series of patients referred to the Knee Clinic in a University Teaching Hospital. We do not claim that they represent a typical adult population. We do not have clinical diagnoses or pain or function scores on these patients and we accept that there is often a difference between the radiological findings of knee osteoarthritis and a patient’s symptoms. The mean interval between X-ray and MRI scans was 3.65 months (SD 3.21, range 0.03–11.87). We did not have sequential scans to allow an assessment of vascular marks changing with time but hope to study that progression in future. A further limitation is our inability at this stage to prove that the MRI marks are blood vessels. However, a literature search failed to identify any other description of these marks on MRI scans and radiological opinion was that they fulfil the MRI criteria for veins or vessels.

In conclusion, our study is the first to describe subchondral vascular marks on MRI scans and to identify an inverse relationship between those marks and the presence and distribution of knee osteoarthritis. Although cause and effect are not established, there is clearly an association. The loss of marks on MRI may be a means of diagnosing or predicting the development of osteoarthritis more accurately than by plain X-ray alone. Future work will involve a review of sequential pairs of X-rays and MRI scans and a histological analysis of the subchondral plane for vessels.