Knee Shape, Leg Alignment and Osteoarthritis: A UK Biobank Study of 19,053 People

Osteoarthritis and Cartilage Open · 2025 · Beynon RA, Alomar F, Saunders FR, Ebsim R, Faber BG, Jung M, Gregory JS, Lindner C, Abram SGF, Aspden RM, Harvey NC, Cootes T, Tobias JH

Key points

  • This study, which I contributed to, used DXA scans from 19,053 UK Biobank participants to examine how knee shape relates to osteoarthritis seen on the scan.
  • It was cross-sectional: everything was measured at a single visit, so it shows associations, not cause.
  • Higher grades of osteoarthritis went with a progressively more bow-legged (varus) knee shape.
  • For the shape pattern with the strongest link, an increase of one standard deviation carried 2.78 times the odds of grade 3 to 4 osteoarthritis.
  • After allowing for whole-leg alignment, these links were weaker, but some differences in shape remained.
  • No treatment was tested, and the results do not predict what will happen to an individual knee.

What Was This Study About?

A bow-legged (varus) leg puts more stress on the inner (medial) side of the knee, and a knock-kneed (valgus) leg more on the outer (lateral) side. Alignment is known to be related to how severe knee osteoarthritis is and how it progresses. It has been less clear whether other features of bone shape also matter.

To look at this, the research team (I was one of thirteen authors) used UK Biobank, a long-term study of around 500,000 people across the UK who were aged 40 to 69 when they joined. Some had DXA scans of the knees and the whole body. A DXA scan (dual-energy X-ray absorptiometry) is a low-dose X-ray scan, normally used to measure bone density. The whole-body scan shows the hip, knee and ankle together, so the alignment of the leg can be measured.

Three measurements were taken for each person:

  • Knee shape. The outline of the bones around the right knee was marked with 129 points. A technique called statistical shape modelling turns these outlines into “shape modes”: the main patterns in which knee shape differs from one person to the next. Each person receives a score for each mode, showing how far their knee is from the average. The first ten modes, accounting for 79.5% of the variation in shape, were analysed.
  • Radiographic osteoarthritis. This is osteoarthritis judged from an image alone. It was graded from 0 (none) to 4, based on the size of bony spurs (osteophytes) and narrowing of the joint space on the inner side of the knee.
  • Leg alignment. The hip-knee-ankle (HKA) angle, a commonly used measure, was calculated from the whole-body scan.

The team then tested whether each shape mode was associated with each grade of osteoarthritis, compared with grade 0, taking account of age, sex, height, weight and ethnicity. A further analysis added the HKA angle, to see whether osteoarthritis was linked to features of shape other than alignment. The analysis was cross-sectional: everything was measured at one visit.

What Did We Find?

Who was included. The 19,053 people analysed had an average age of 63.7 years; 52% were women and 96.8% were white. Most had no osteoarthritis on the scan: 80.7% were grade 0, 14.6% grade 1, 3.6% grade 2 and 1.2% (220 people) grade 3 or 4, which were analysed together.

Shape and osteoarthritis grade. In the first analysis, several shape modes were associated with osteoarthritis, more strongly at higher grades. The results are odds ratios for each increase of one standard deviation (a statistical measure of distance from the average) in a shape mode. A value above 1 means higher odds of being in that grade than in grade 0; below 1 means lower odds. For mode 8, the odds ratio was 1.19 for grade 1, 1.52 for grade 2 and 2.78 for grades 3 to 4. Mode 9 ran the other way, at 0.78, 0.71 and 0.58.

The combined picture. Combining the ten modes into one outline for each grade showed a progressive shift towards bow-legged (varus) alignment as the grade increased. At grades 3 to 4 there was also a narrower joint space on the inner side of the knee and a wider lower end of the thigh bone.

After allowing for leg alignment. With the HKA angle added, the bow-legged appearance was considerably less marked, and the narrower joint space and wider thigh bone were no longer apparent. Alignment accounted for much of the link between knee shape and osteoarthritis, but not all of it. Modes 8 and 9 remained associated with every grade, with smaller effects, and the outline for grades 3 to 4 now showed a kneecap sitting further towards the outer side of the knee and a narrower upper end of the shin bone.

Men and women. A few individual modes behaved differently in men and women, but the composite outlines were similar in both.

Why Does This Matter for Patients?

The main finding is a familiar one: more advanced osteoarthritis went with a more bow-legged knee, consistent with earlier research. The study cannot say which comes first. Bow-legged alignment may predispose a knee to osteoarthritis or aggravate it by concentrating load on the inner side of the joint. It may also be a consequence of the joint space narrowing on that side.

What is new is that some differences in shape remained once alignment had been accounted for. In the paper we say these may be of significance and justify further study.

No treatment was tested, and the paper makes no recommendation about how knee osteoarthritis should be assessed or treated. A knee osteotomy is an operation that corrects the alignment of the leg, but these results say nothing about who would benefit from one.

What Could This Study Not Tell Us?

Because everything was measured at one point in time, the study cannot show that a particular knee shape causes osteoarthritis, or predict who will develop it. Osteoarthritis was graded from the image alone, and what a scan shows does not always match how a knee feels.

Joint space narrowing was part of the definition of osteoarthritis, which may itself account for the link between osteoarthritis and a narrower joint space.

DXA scans are taken lying down, not standing, which typically gives a larger measurement of the joint space, and to our knowledge no study has compared knee DXA scans directly with standing X-rays. They are two-dimensional, give a limited view of osteophytes and can be distorted by small changes in positioning. In the paper we note that the differences in the width of the thigh and shin bones may be effects of projection onto a flat image, although MRI studies have reported a similar widening of the thigh bone in osteoarthritic knees.

UK Biobank participants are not typical of the wider population: 96.8% of those in this study were white, and UK Biobank as a whole has a lower death rate than the general population, so the findings may not apply to everyone.

Frequently asked questions

Does being bow-legged cause knee arthritis?

This study cannot answer that. Among 19,053 UK Biobank participants, higher grades of osteoarthritis on DXA scans went with a more bow-legged (varus) knee shape, but everything was measured at a single visit. Bow-legged alignment may contribute to osteoarthritis, and it may also be a consequence of the joint space narrowing on the inner side of the knee.

Is the shape of the knee linked to osteoarthritis?

In this study of 19,053 people, several patterns of knee shape were associated with osteoarthritis seen on DXA scans, more strongly at higher grades. For one pattern, an increase of one standard deviation carried 2.78 times the odds of grade 3 to 4 osteoarthritis. Leg alignment accounted for much of this. These are averages across a large group and do not predict what will happen to an individual knee.

What is radiographic knee osteoarthritis?

Radiographic osteoarthritis is osteoarthritis judged from an image alone. In this study it was graded from 0 (none) to 4 on DXA scans, based on bony spurs (osteophytes) and narrowing of the joint space on the inner side of the knee. Of 19,053 people, 80.7% were grade 0 and 1.2% were grade 3 or 4.

Can a DXA (bone density) scan show knee arthritis?

In this research, knee DXA scans were used to grade osteoarthritis and measure knee shape in 19,053 people. DXA scans are taken lying down, which typically gives a larger measurement of the joint space than a standing X-ray, and the paper reports that no study known to the authors has compared the two directly.

Read the Full Paper

This paper is published open access in Osteoarthritis and Cartilage Open under a Creative Commons licence (CC BY 4.0). You can read the full text below, or on the journal website or on PubMed Central.

Your browser does not support embedded PDFs. Download the PDF or view on the journal website.

Full citation: Beynon RA, Alomar F, Saunders FR, Ebsim R, Faber BG, Jung M, Gregory JS, Lindner C, Abram SGF, Aspden RM, Harvey NC, Cootes T, Tobias JH. Use of dual-energy X-ray absorptiometry to evaluate variation in bone shape and alignment associated with radiographic knee osteoarthritis: Findings from a study of 19,053 individuals in UK Biobank. Osteoarthritis and Cartilage Open. 2025;7(4):100667. doi:10.1016/j.ocarto.2025.100667

This summary is written for general information and explains the research in patient-friendly terms. It does not replace individual medical advice. If you have questions about knee osteoarthritis or your knee, please get in touch to arrange a consultation.

← Back to Research, Guidelines & Articles