How long does it take to develop PTA? This is an area where conventional wisdom has been challenged in recent years. It used to be said that if arthritis was going to develop, some evidence would be apparent within 2 years. At 2 years post-injury, confident opinions about future arthritis were then expressed.
This was based on papers published 40-60 years ago, when treatment was less rigorous regarding the maintenance of anatomic reduction. Follow-up in these studies was also relatively short. In Lindsjo’s thorough 1981 study, the mean follow-up was 3.7 years, the maximum follow-up was six years (Lindsjo, (1981) Operative Treatment of Ankle Fractures, Acta Orthopaedica Scandinavica, 52:sup189, 1-131, DOI: 10.3109/ort.1981.52.suppl-189.01).
Of course, an accelerated course of arthritis (following severe injury, imperfect reduction, or infection) may be detectable in this 2 year timescale. However, after anatomic reduction one would expect the timescale of PTA to be longer. A further point is that papers based on a short follow-up will under-estimate the lifetime incidence of PTA, if the latency is in fact longer than the follow-up.
1985, Phillips et al. ‘A Prospective, Randomized Study of the Management of Severe Ankle Fractures’. J Bone Joint Surg Am 67: 67-78. n=138. After acceptable initial closed reduction, patients randomised to closed or open treatment by AO/ASIF methods. Follow-up in this urban trauma centre study was only 51% to average 3.5 years, so this was a relatively short term study which may under-estimate the eventual incidence of arthritis. These authors confirmed that a medial malleolar fracture was associated with worse outcome. They found that the only radiological measurement that was predictive of arthritis was the difference in the talocrural angle between the fractured side and the uninjured side. The talocrural angle is obtained by drawing a line along the distal tibial articular surface, then a perpendicular to that. A line is then drawn between the tips of the malleoli. The angle is the superior/medial angle formed by the intersection of the latter two lines. The difference between sides should be less than or equal to 5 degrees, which was the smallest difference that could be reliably estimated by these authors. The normal figure is 83 degrees +/- 4 degrees. A talocrural angle difference of more than 5 degrees was significantly associated with the development of arthritis (p value less than 0.01). This result supports (1) the importance of maintaining fibular length, and (2) that the quality of reduction is important. However, many of the methods to assess this (listed in the publication) do not reliably distinguish between those who will develop arthritis and those who won’t (at least over the medium term).
Stufkens et al, ‘Long-term outcome after 1822 operatively treated ankle fractures: A systematic review of the literature’. Injury, Int. J. Care Injured 42 (2011) 119–127. This study illustrates the lack of consistency in evaluation between series. However, unsurprisingly, it did show that the worse the reduction, the worse the outcome. There wasn’t much difference in outcome between Weber A & B, but Weber C was worse. Lauge-Hansen
SER-2 type fractures have a superior long-term outcome compared to SER-4 type fractures. Fractures with cartilage lesions have a considerably worse prognosis, (OR =5). Only 79.3% of the optimally reduced fractures show good to excellent long-term outcome. Comment: Once again, with a follow-up of ‘at least 4 years’, the lifetime risks of PTA are likely to be under-estimated. (Useful list of references).
Horisberger et al, ‘Post-traumatic ankle osteoarthritis after ankle-related fractures’. J Orthop Trauma, 2009 Jan; 23(1): 60-67. The latency period between injury and end-stage ankle PTA was 20.9 years. A negative correlation was observed between severity of injury and latency time for pilon fractures. Complications during healing were associated with shorter latency. Age at injury correlated negatively with latency time.
Lubbeke et al, ‘Risk factors for post-traumatic osteoarthritis of the ankle: an eighteen year follow-up study’. International Orthopaedics (SICOT) (2012) 36:1403–1410 DOI 10.1007/s00264-011-1472-7. This study looked at 373 fractures (372 patients; 9% Weber A, 58% Weber B, 33% Weber C) that had been operated upon. The mean age at operation was 42.9 years. There were 102 patients seen at follow-up (mean follow-up 17.9 years). Those not available for review did not differ in demographics and fracture type from those reviewed. Advanced radiographic OA was present in 37 patients (36.3%). Significant risk factors were: Weber C fracture, associated medial malleolar fracture, fracture-dislocation, increasing body mass index, age 30 years or more and length of time since surgery. Advanced radiographic OA was common 12–22 years after malleolar fracture. The probability of developing post-traumatic OA among patients having three or more risk factors was 60–70%.
It is interesting that, according to these authors, fracture-dislocation is a risk factor. This could be just a marker of a more severe injury, but it could also be a surrogate for cartilage damage. That would support the conclusion of Stufkens et al about cartilage lesions (see above). The rough surface of a fractured malleolus acts like a pan-scrubber on the soft cartilage of the talus as it is dragged across it. I have seen those lesions many times.
2015, Verhage et al. ‘Long-term functional and radiographic
outcomes in 243 operated ankle fractures’. Journal of Foot and Ankle Research (2015) 8:45. DOI 10.1186/s13047-015-0098-1. These authors, like Lubbeke et al, found that fracture of the medial malleolus was important. They concluded that long-term functional outcome was strongly associated with medial malleolar fractures, isolated or as part of bi- or trimalleolar fractures. More cases of osteoarthritis are found in trimalleolar fractures.
1999 Meyer-Wolbert et al, Chirurg (1999) 70: 1323-1329. ‘Die prognostische Bedeutung verletzter Anteile bei Sprunggelenkfrakturen’. These authors said: ‘Injuries of the medial ankle side or the posterior tibial margin do not necessarily predict a negative
outcome in ankle fractures, as shown in this comprehensive retrospective study on 342 consecutive patients. A negative predictive value results from the combination of different injured ankle components. Clinical and radiological results correlate well with each other and with the severity of the fracture. They deteriorate according to the number of injured ankle components’.
The posterior ‘malleolus’ fracture (meaning fracture of the posterior lip of the tibia at the ankle joint) is held by a number of authors to be a predictor of osteoarthritis. It is not clear whether this risk is completely averted by accurate reduction & fixation, but one would expect some effect.
These authors include:
1981, Lindsjo: (Lindsjo, (1981) Operative Treatment of Ankle Fractures, Acta Orthopaedica Scandinavica, 52:sup189, 1-131, DOI: 10.3109/ort.1981.52.suppl-189.01). This is more of a thesis than a paper, available to download, and worth reading.
1985, Bauer. (Mats Bauer, Kjell Jonsson & Bo Nilsson (1985) Thirty-year follow-up of ankle fractures, Acta Orthopaedica Scandinavica, 56:2, 103-106, DOI: 10.3109/17453678508994329. https://doi.org/10.3109/17453678508994329). n=143; average follow-up was in fact 29 years. This group included ‘all-comers’, even minor Lauge-Hansen grades. The outcomes therefore were predictably better than other series in which worse LH grades or operative groups are reviewed. 82% had no radiographic signs of arthrosis; 83% were free of symptoms. The most common fracture, supination-eversion Stage II (49 cases), gave rise to minimal signs of arthrosis in only one patient, who had moderate symptoms.
2002, Langenhuijsen et al, J Trauma. 2002;53:55–60. ‘Results of Ankle Fractures with Involvement of the Posterior Tibial Margin’. n= 57 reviewed out of 85 at 6.9 years. Modified Weber endscore used for assessment, in which 25% relates to OA features. These authors concluded that the involvement of the articular surface ranged from 8% to 55%. Size or fixation of the fragment did not influence prognosis. Joint congruity in fragments > 10% of the articular surface was a significant factor influencing prognosis. Overall, the modified Weber protocol result was excellent in 10%, good in 15%, fair in 25%, and poor in 50% of patients. Joint congruity with or without fixation was a significant factor influencing prognosis. Congruity should be achieved for fragments > 10% of the tibial articular surface. The authors felt the modified Weber score was ‘too strict”, crediting only nearly perfect results.
2020, Blom et al on behalf of EF3X-trial Study Group. ‘Posterior Malleolar Fractures. Predictors of Outcome.’ Bone Joint J 2020:102-B(9):1229-1241
Posterior Malleolar Ankle Fractures (PMAF) of a rotational type can be classified into 3 types on CT:
Haraguchi Type 1: Large Posterolateral-oblique
Haraguchi Type 2: Two-part posteromedial and posterolateral
Haraguchi Type 3: avulsion
n=70. Assessment = Foot & Ankle Outcome Score (FAOS).
Hamaguchi Type 2 PMAFs have worse outcomes as assessed by 2 year FAOS than Type 1 & Type 3.
Type or morphology of fragments, not size, seems to be the most important factor determining outcome. Whether or not to fix PMAFs is likely to depend on a better understanding of the benefits of fixing (or not) the distinct morphological subtypes.
OA can be a radiological or clinical definition, but the need for a fusion or replacement is definitive as an outcome following these injuries.
A publication in the J. Orthop Trauma in Jan 2020 by Axelrod et al (PMID: 31851114) looked at the risk of fusion/arthroplasty in a large cohort of operatively treated (n:about 44000) and nonoperatively treated ankle fractures (n: about 88000), and compared these with the population. The inclusion criteria were “rotational ankle fractures”, though this does include
Their summary findings were that the risk of needing such surgery was about 0.65% for an operatively treated ankle fracture, and about 0.17% for nonoperatively treated fractures, with about a 20x increased risk compared to the population in general.
The authors note that they could not explore more recently noticed subtypes of such injuries, could not quantify if a poorer reduction was associated with the increased risk of needing fusion, and that this percentage may under-represent the actual incidence of PTOA.
However, the strengths include the sheer numbers in each group and the robustness of the outcome measure.
Many thanks for the comment Badri. This paper uses a most interesting approach, and as you say its strength is the big numbers. Its weaknesses appear to be (unless I’m reading it wrong): (1) It depends on the unknown accuracy of coding of primary & secondary operations, and linkage between the first & second. Having been accused of performing cholecystectomy as a T&O consultant, on the basis of coding, I’m naturally a bit sceptical of the accuracy & linkage of coding. (2) Anyone who moves in or out of state is not included. (3) The patients between 1994 and 2011 were included (17 years), but if Horisberger’s paper is correct and the usual latency between injury and end-stage PTA is 20 years, then the majority did not have a long enough interval to develop end-stage PTA.Thus the paper could be under-estimating the lifetime risk.
Whilst the incidence of arthrodesis/arthroplasty surgery appears to be a hard endpoint (if coding & linkage problems were overcome) it will of course omit those who had a significant amount of PTA but not enough for surgery; those who had died with their PTA; those who had PTA severe enough for surgery but who were unfit. Also I think the data has to be looked at in the context of the epidemiology of ankle fracture. In most studies this shows a different pattern for M&F, and two age peaks (young adulthood and old age). Those in the second age peak may not have time to develop PTA, or may be immobile for other reasons or unfit for major ankle surgery. So, for all the big numbers, I think this approach will capture only a part of the PTA problem.