Management of weight-bearing bones totally affected by FD is challenging, and it may lead to patient disability when it is inappropriate, as in the patients of our cohort before IN [3, 6, 7, 12, 17]. According to our results, IN—indicated at present as the best primary treatment in long bones extensively affected by FD [3, 9,10,11,12,13,14,15,16,17, 26, 27]—appears to also be ideal as a salvage procedure after improper treatments.

Published expert opinions, case reports and small series with a short-term follow-up on femoral and tibial primary IN in PFD/MAS are prevalent [3, 9,10,11,12,13,14,15,16, 26]. We operated on 48 femurs and tibias of 24 patients, with an average follow-up of 9 years. As far as we know, this is one of the largest series of lower limb IN with a long-term follow-up, albeit performed after previous unsatisfactory treatments.

Our patients had had various treatments with poor results. Therefore, we can confirm that non-operative orthopedic procedures as well as femoral diaphyseal nailing without cervical stabilization must be abandoned [3, 4, 6, 9, 10, 12, 15,16,17]. Since peripheral plating frequently fails [3, 9,10,11,12,13,14,15,16,17, 26, 27], its indication should be very limited [27], although for several authors, peripheral plating still represents the preferred treatment in PFD/MAS patients with lower limb involvement [18,19,20,21,22,23,24,25].

The Zickel nail was the first cervicodiaphyseal femoral device used in PFD to prevent coxa vara thanks to its femoral neck component. On the basis of its positive use [9], validated by a follow-up of more than 20 years [3], we employed the new generation of cervicodiaphyseal nails, which also include adolescent nails for smaller femurs [30]. In the tibia, regular interlocking tibial nails were used, whereas smaller tibias were nailed with adult interlocking humeral nails.

We were able to fix all the femoral and tibial deformities except for 4 femoral type 6 deformities in which a residual varus angulation of the femoral shaft was still present after IN and remained unchanged at follow-up. Unfortunately, in those cases, the extreme softness of the fibrodysplastic bone surrounding the medullary canal drilled into the bone segments after the osteotomy did not allow the nail to maintain a straight alignment of all the femoral shaft osteotomy segments.

Only 3 validated scales are available to assess PFD patients [11, 31, 32]. Enneking’s scale has been specifically conceived for patients operated on for bone tumors [31]. Since PFD is a benign condition, we adopted the Guille et al. scale [33] as modified by Jung et al. [11]. According to the Jung scoring system—made up of 5 basic clinical parameters—almost 83% of our patients obtained satisfactory long-term results, and the clinical score for the 20 nonfractured patients significantly improved at follow-up with respect to their score before IN, which we performed after previous improper treatments.

A high incidence of fractures up to 30 years of age has been reported in MAS [5]. Therefore, fracture prevention was another important goal achieved by IN, since 83% of our patients were younger than 30 years at IN, and the 2 femoral fractures after IN could not be prevented because one occurred below the tip of the nail and the other was caused by a trauma so intense that it bent the nail cervical blade.

Both the extent of lower limb involvement and the quality of FD bone affected the result of treatment, since monomelic MAS patients fared better than bimelic ones—although the difference was not significant—and monomelic PFD patients fared significantly better than those with MAS. In the latter, the hormonal influence worsens the weakness of FD bone, as already reported [3, 5,6,7,8, 12, 17]. In this context, the loss of neck–shaft angle correction observed in 11 femurs of our MAS patients at follow-up, regardless of the support of the nail’s cervical component, may be explained by the extreme bone weakness of the upper femur.

Only 1 MAS patient had a recurrence of a type 6 femoral deformity, but she had been jogging 3 times a week for 15 years since the IN. We believe that sporting activities should be carefully selected in MAS patients with severe lower limb involvement, since IN per se is not sufficient to prevent deformity recurrence when the limb is overloaded.

Coxa brevis responsible for ROM limitation was present in several hips. It might have been caused by repeated surgery that stunted femoral neck growth, as happened in the patient illustrated in Fig. 2.

Four MAS patients ended up with an unsatisfactory result. In 3 patients with extensive pelvic involvement, one hemi-pelvis gradually collapsed, leading to a marked LLD that was compensated for with a shoe lift and an aid for walking. Severe pelvic involvement by FD was the worst risk factor for a poor result in our cohort but, as far as we know, no surgical solution has been described for this condition. The fourth patient developed severe hip osteoarthritis at 30 years of age [32]. However, surgery was not performed because the result of total hip replacement is unpredictable in PFD [34].

Lower limb length discrepancy (LLD) correction was a problem in our cohort, as already reported [3, 12, 17]. We performed shortening osteotomy, but residual LLD was still present in two-thirds of our patients, though only 8 wore a shoe lift, including the 3 with a hemi-pelvis collapse. Bone lengthening has been reported in PFD when the cortical bones are thick enough to withstand distraction [35]. Unfortunately, our cases were totally affected by FD and their cortical bones were very thin, thus not allowing lengthening [3, 17].

Knee height asymmetry was present in one-third of our patients. It was caused either by the spontaneous overgrowth of the femur or tibia—often observed in the same limb of PFD/MAS patients [2, 3, 6, 11, 12]—or by previous treatments that stunted their growth. We tried to reduce the asymmetry by performing a shortening osteotomy of the longer bone segment in cases with LLD. However, even 5 cm of residual knee asymmetry did not cause any significant limitation on the usual activities of daily living.

We had a high incidence of complications (21%). We were very surprised by the low complication rate in previous studies on primary lower limb IN; this was probably due to the small number of cases and/or to the shortness of the follow-up [4, 9,10,11, 13, 14, 18,19,20,21,22,23,24,25]. Zhang et al. [36] deserve particular mention for the absence of reported complications in 65 patients treated with IN during a mean follow-up of 6 years. However, we should also bear in mind that previous repeated treatments—both non-operative and surgical—might have had a negative influence on FD bone quality in several cases, causing mainly either non-union or delayed union after IN. None of the cases suffered an infection, as also previously reported [9,10,11,12,13,14,15,16,17,18,19,