Although all implants used in this study were withdrawn from the market about 20 years before, the longitudinal clinical outcomes over decades will help to better understand potential factors leading to implant failure or complications and assess the safe and predictable use of dental implant. Our analyses revealed a 25-year cumulative survival rate of 89.8% after the prosthesis setting, which seems comparable to the result of a recent study . Although approx. 62% of the patients and implants in our original cohort dropped out during the follow-up period, according to Table 1, a majority of patients who underwent implant treatment were middle- and old-aged (82.6% of the patients were 40–69 years old) and thus some of the patients could not continue maintenance for varying reasons over a 25-year follow-up. The other 38% of the patients are healthy and likely to visit their dentists for maintenance, and they were included in the 25-year cumulative survival rate . Therefore, the true long-term survival rate might have been lower than we reported herein due to bias from the patients who dropped out.
In addition, only four implants of one patient were inserted into a re-constructed site from the iliac bone; no other bone augmentation procedure such as bone graft, guided bone regeneration (GBR), and sinus floor elevation were conducted. The principle of guided tissue regeneration (GTR) was introduced in 1982 , and GBR was introduced in 1988 . The technique of sinus floor elevation was initially introduced in 1980 . These complex augmentation procedures had not been common at that time , especially in private practices in Japan, and thus, they were not used for any of the patients in the present study. There is thus some degree of bias regarding the numbers of implant and the lengths of the implants according to the implant position. The number of implants applied to the maxilla anterior region was only two, since getting the esthetic result with implant prostheses was uncertain in those days. And the number of implants under 10 mm long was greater at posterior sites compared to anterior sites due to the sinus and inferior alveolar nerve.
Peri-implantitis is the major reason for late failure [13, 14]. The consensus report of the Sixth European Workshop on Periodontology described peri-implant mucositis in approx. 80% of subjects restored with implant, and peri-implantitis in 28–56% of subjects . In the present study, the cumulative incidence of peri-implantitis was 9.5, 15.3, 21.0, and 27.9% at 5, 10, 15, and 25 years after the prosthesis setting, respectively. Derks and Tomasi reported a positive relationship between the incidence of peri-implantitis and the mean function time by performing a meta-regression analysis of a systematic review , whereas the current cumulative result shown in Fig. 3 may represent the time course of the peri-implantitis incidence. Interestingly, the incidence of the peri-implantitis increased gradually with time; the rate of increase was approx. 1–1.5% per year.
Many potential factors associated with the incidence of peri-implantitis were reported [17, 18]. In the present study, the gender, implant type, and width of keratinized mucosa were identified as risk factors. Regarding gender, Koldsland et al. also reported a male population with overt peri-implantitis , whereas Attard and Zarb reported that women experienced more peri-implant bone loss than men . Other studies and reviews reported that gender had no effect on peri-implantitis [21, 22]. Some other gender-related factors might affect the results.
Regarding implant type, a difference between the S-types and the TPS-types is whether the existence of an abutment connection or not. The TPS-types are one-piece implants, and the S-types are two-piece but one-stage implants. Duda et al. reported that one-piece implants showed more marginal bone loss than two-piece implants . In addition, a TPS surface is classified as “rough” surface when the surface roughness is more than 2 μm (Sa > 2 μm) . Teughels et al. reported that a transmucosal implant surface with higher surface roughness facilitates biofilm formation  and thus TPS-type implants showed a higher incidence of peri-implantitis compared to the S-type.
Regarding the width of keratinized mucosa, many studies and a review have indicated that the presence of a sufficient width of keratinized mucosa is necessary for maintaining healthy peri-implants [26,27,28,29]. In the present study, when 2 mm of keratinized mucosa was used as the adequate width, the p value was 0.053 (data not shown). This also showed the tendency of the availability of keratinized mucosa around implants, and it may indicate that at least 2 mm of keratinized mucosa is preferable for the long-term success and survival of implants.
Our analysis showed that 16 of 223 implants were lost during the observation period. Among the six factors examined, only the implant position affected the cumulative implant survival rate and the main reason for implant failure was peri-implantitis (14/16 failed implants). However, the implant position did not affect the incidence of peri-implantitis. Compared to the mandible, the bone quality of the maxilla is lower  and the loading force is tilted to the implant axis. These factors might have acted as an exacerbating factor of peri-implantitis, resulting in the lower survival rate of the implants in the maxilla compared to the mandible.
Prosthetic complications occur due to the accumulation of mechanical damage to the implant, implant components, and supra-structures, resulting in the need for repairs and reconstructions of the implant prostheses, which may require time-consuming procedures and additional financial resources. The present investigation was a retrospective and multicenter study, and there were many differences in design patterns, materials, connections, and the attachment of supra-structures. It was therefore difficult to subdivide and review the factors that may affect the prosthetic survival rate, and only gender and type of prosthesis could be analyzed in this study.
The implant-supported fixed prostheses showed the highest complication-free survival rate in our study. It was reported that the veneering material’s chipping/fracture is the most common type of prosthetic complication for fixed prostheses [31, 32]. Pjetursson et al. reported that veneer fracture was observed in 13.5% of fixed prostheses after at least 5-year functioning . In the present study, approx. 76% of the fixed prostheses were not veneered (metal occlusal surface), resulting in the lower complication rate after 25 years of functioning.
We also observed that the tooth-implant-supported prostheses had a lower complication-free rate than implant-supported fixed prostheses due to caries, periodontitis, or the root fracture of abutment teeth. Lang et al. reported that the survival rates of tooth implant-supported fixed partial dentures were 94.1% after 5 years and 77.8% after 10 years of functioning , and these results were almost the same as ours (93.9% after 5 years’ and 77.2% after 10 years’ functioning). Taking our results and those of Lang et al. into account, it appears that prosthetic complications of tooth implant-supported prostheses start arising after 7 years post-setting and then increase with time.
The implant-supported overdentures showed the lowest complication-free rate among the three implant types in the present study, due to the wear or fracture of artificial teeth, attachment fracture, and relines. Compared to another retrospective study of conventional complete dentures (without implant support) , our complication-free rate was higher and there was a difference in terms of the incidence of artificial tooth problems. That study showed a < 10% of incidence of artificial tooth problems during the first 5 years post-setting. The rigid support provided by an implant might have enhanced the loss, wear, and fracture of artificial teeth in our patients.