No major adverse events were observed during procedure or follow-up period. 87.6 % patients of PVP/PKP group returned to their normal lives 1 week after operation and 59.2 % patients of conservative group exercise to their regular level 2 months later. In total, 31 of 290 (10.7 %) patients returned with 42 new OVCFs (Table 1). 14 individuals (5 males, 9 females) were initially treated by surgical technique (PVP 9, PKP 5) and 17 ones (7 males, 10 females) were from conservative group. 25 patients were subjected to repeated PVP/PKP and the other six patients were successfully managed by conservative care. Of these newly documented fractures, 14 (34.2 %) involved adjacent segments, whereas 25 (59.5 %) occurred at distant levels, and another three (7.3 %) were the same.
In terms of frequency of recompression, we selected a method similar to that of Dr Johannes et al. . Fourteen (8.28 %) of these 169 patients developed 18 new OVCFs, three (21.4 %) of the 14 patients had a third compression fracture and a third PVP. Excluding the cervical spine as well as TH1–TH2 segments due to low risk of OVCFs, the other 15 levels (TH3–L5) were considered to occur new fracture(s) potentially. In the PVP/PKP group, 217 bodies were intervened initially and 18 (0.71 %) of the whole 2,535 levels (169 patients × 15 vertebrae) developed new OVCFs, nine of them belonged to adjacent segments, the other nine belonged to distant segments, none of the pretreated segments recollapsed. Of these 2,318 (2,535–217) vertebral bodies, 360 (15.5 %) were adjacent to pretreated segments, while 1,958 levels came from the distant. Therefore, the percentages of new adjacent/distant OVCFs were 2.5 % (9/360) and 0.46 % (9/1,958) for the second compression fractures (χ
2 = 16.4, p = 0.00). For the third new OVCFs, the data were 5.12 and 1.53 %, respectively (p = 0.239). Adjacent fractures were more prevalent than distant ones for the second fractures; however, the differences had not reached statistical significance for the third-time fractures. Of the 18 new fracture segments, ten new fracture segments were from levels above the pretreated segments, the other eight were from levels below the pretreated segments, and no refracture of pretreated segments was detected. The distance from the initial fracture segment in distant group was 2, 2, 2, 2, 3, 3, 4, 7, 7 level(s), respectively. Calculated by patient number, six patients had new adjacent fractures and the other six was from nonadjacent cluster, with two patients experienced both adjacent and non-adjacent fractures.
The mean age of the 155 patients was 70.9 years, a little younger than the refracture unit (73.1 ± 8.93, t = 1.974, p = 0.43). Males comprised 31.6 % (51 cases), and in the refracture group it was 35.7 % (5 cases, p = 1.00).
The distribution of original and new OVCFs is shown in Table 2. New fracture(s) occurred mostly in the thoracic and thoracolumbar segment compared with the previous fracture(s), but difference was not statistically significant (t = 2.11, p = 1.00).
The time interval was 8.95 ± 7.34 months for adjacent group and 10.75 ± 8.68 months for nonadjacent group. It seems that adjacent new OVCFs occurred earlier than nonadjacent levels; however, we did not find a statistically significant difference (t = 1.548, p > 0.05). Of the three patients that experienced third OVCFs,the symptom-free interval after a second compression fracture was 6, 2 months for adjacent segments and 15 months for a patient in the same treated segment. Owing to limited number of third fracture cases, we could not compare the data effectively (Fig. 2).
The mean number of preexisting fractures in single time fracture group was 1.36 (1.36 ± 0.75, 95 %CI, 1.24–1.47) and in repeated fracture group, 1.57 (1.57 ± 1.11, 95 %CI, 0.92–2.21). As can be seen from the above, baseline number of vertebra fracture is not a predictor for new fractures (t = 1.972, p > 0.05). The amount of injected cement was 3.95 ± 1.43 (range 1.5–9 mL) in single time fracture group and 3.26 ± 1.54 (range 1–6 mL) in the repeated fracture group. There was no positive association between the quantity and recollapse of new segment (t = 1.4651, p > 0.05).
PMMA leakage was observed in four cases (1.77 %) after the first incident: two into the disk, one into the lateral blood plexus and one with anterior extravasation going up the above level (Fig. 1b T11). None of the leaks exuded into neural canal or caused neurologic symptoms. None of the patients developed refracture.
During the same follow-up period, 17 (14.0 %) of the 121 patients went through 24 (24/1, 815,1.32 %) new OVCFs, two (11.7 %) developed a third compression fracture and a third PKP. Five segments belonged to adjacent section, 16 from nonadjacent ones, and the remaining 3 from the same previous fracture segments. Using the same calculating method with PVP/PKP group, the proportion of adjacent new fractures (1.91 %, 5/261) is also higher than nonadjacent fractures (1.13 %, 16/1,408), but did not reach statistical significance (χ
2 = 0.519, p = 0.471). Eight segments occurred above pretreated levels; thirteen from below sites, while 3 were from the same segments.
As the number of third fracture was limited, it was difficult to monitor the relative risk of new fractures sufficiently. With the accretion of distance from the original fracture level, recompression incidence decreased gradually. The mean time interval was 22.4 ± 7.99 months (range 0.67–72 months). It seems that fracture occurrence in adjacent segments was more delayed than nonadjacent segments (28 ± 3.46 versus 18.8 ± 14.3 months);however, the data had not reached statistical significance.(t = 1.361, p > 0.05). The original and new OVCFs group also did not differ significantly in age, gender distribution, location of original fracture segments, and number of preexisting fractures.
Comparison between PVP/PKP group and conservative group
We exclusively compared the data of second refractures due to only a minority of third recompressions. From our data, incidence of recompression in the conservative group is higher than that in the PVP/PKP group, with 14.05 % compared with 8.28 % concerning patient proportion and 1.32 % compared with 0.71 % regarding refracture number. The relative risk is 1.69, but no statistical significance has been reached (χ
2 = 2.455, p = 0.117). Both groups indicate a higher incidence of adjacent recompression; however, the relative risk in PVP/PKP group (2.5 %, 9/360) is not significantly higher than that in conservative group (1.91 %, 5/261, p = 0.628).
Most of the second fractures in the PVP/PKP group occurred within the first year (71.42 %) compared with less than half of the recompression in the conservative group (47.05 %),and the mean interval was also much shorter in PVP/PKP group (9.72 ± 17.8 months) than that in the conservative group (22.48 ± 7.99 months, p = 0.017). Adjacent and nonadjacent new OVCFs of PVP/PKP group also occurred remarkably sooner than in the counterparts of conservative group. No major differences were observed in other parameters as presented in Table 3.
After the initial treatment, a total of 27 sandwich situations were created in both groups, no sandwich fracture was observed in our follow-up period. Time interval distinction between above and below levels after PMMA injection also did not reach statistical significance (10.71 ± 7.74 versus 8.5 ± 7.53 months, p = 0.62), so did conservative therapy (25.6 ± 13.3 versus 20.0 ± 11.7 months, p = 0.45).