gabri.grusovin@tiscali.it; r_pistilli@libero.it; fguali@tin.it; leest007@hanmail.net;

anyridge@imegagen.com; holeonemedical@libero.it; sian.matthews@megagen.org;

h.maghaireh@yahoo.co.uk; yoojaw@hanmail.net; jacopo.buti@manchester.ac.uk

Version 03/03/15

Dental implants with internal versus external connections: 1-year post-loading results

from a pragmatic multicenter randomised controlled trial

Marco Esposito, Hassan Maghaireh, Roberto Pistilli, Maria Gabriella Grusovin, Sang Taek

Lee, Federico Gualini, Jungtaek Yoo, Jacopo Buti

Marco Esposito, DDS, PhD, free lance researcher and Associated Professor, Department of

Biomaterials, The Sahlgrenska Academy at Göteborg University, Sweden

Hassan Maghaireh, BDS, MFDS, MSc, Lecturer in Implant Dentistry, University of

Manchester, UK

Roberto Pistilli, MD, Resident, Oral and Maxillofacial Unit, San Filippo Neri Hospital,

Rome, Italy

Maria Gabriella Grusovin, DDS, private practitoner in Gorizia, Italy

Sang Taek Lee, DDS, private practitioner, Suncheon, South Korea

Federico Gualini, MDS, private practitoner in Lovere and Bergamo, Italy

Jungtaek Yoo, DDS, private practitioner, DaeJeon, South Korea

Jacopo Buti, DDS, PhD, Clinical Lecturer in Periodontology, School of Dentistry, The

University of Manchester, United Kingdom

1

Correspondence to: Dr Marco Esposito, Casella Postale 34, 20862 Arcore (MB), Italy. E-

mail: espositomarco@hotmail.com

Key words: Complication, dental implant, external connection, internal connection

Running title: Internal versus external connection

Abstract

Purpose: To evaluate advantages and disadvantages of identical implants with internal or

external connections.

Materials and Methods: Two-hundred-thirty patients with any type of edentulism (single

tooth, partial and total edentulism) requiring one implant-supported prosthesis were

randomly allocated in two equal groups to receive either implants with external connection

EC) or implants of the same type but with internal connection (IC) (EZ Plus, MegaGen

Implant Co. Lld. Gyeongbuk, South Korea) at 8 centres. Due to slight differences in implant

design/components IC implants were platform switched while EC were not. Patients were

followed for 1 year after initial loading. Outcome measures were: prosthesis/implant failures,

any complication, marginal bone level changes and clinician preference assessed by blinded

outcome assessors.

Results: One-hundred-seventeen patients received 205 EC implants and 113 patients 202 IC

implants. Seven patients dropped-out with 12 EC implants and four patients with 5 IC

implants, but all remaining patients were followed up to 1-year post-loading. Three centres

did not provide any periapical radiographs. Two prostheses supported by EC implants and

one by IC implants failed. Three EC implants failed in three patients versus two IC implants

in one patient. EC implants were affected by nine complications in nine patients versus seven

complications of IC implants in seven patients. There were no statistically significant

2

differences for prosthesis/implant failures and complications between the implant systems.

One year after loading, there were no statistically significant differences in marginal bone

level changes between the two groups (P=0.0629) and both group lost bone from implant

placement in a statistical significant way: 0.98 mm for the EC implants and 0.85 mm for the

IC implants. Five operators had no preference and three preferred IC implants.

Conclusions: Within the limitations given by the difference in neck design and platform

switching between EC and IC implants, preliminary short-term data (1-year post-loading) did

not show any statistically significant differences between the two connection types, therefore

clinicians could choose whatever they prefer.

Conflict of interest statement: This trial was partially funded by MegaGen Implant Co. Lld.

Gyeongbuk, South Korea, the manufacturer of the implants evaluated in this investigation,

however data belonged to the authors and by no means did the manufacturer interfere with

the conduct of the trial or the publication of the results.

INTRODUCTION

Implant-supported prostheses are an effective and reliable treatment for replacing

missing dentition. The success of implant-supported prostheses is mainly based on the

"integration" of dental implants in newly formed bone. This process is generally known as

"osseointegration". Literally thousands of new dental implant design, materials and surface

technologies are continuously developed to further improve the outcome of implant therapy,

many of them claiming superiority over competitors. There are many randomised controlled

trials comparing different dental implant made of various materials and having different

design, and surface characteristics.(2) Most of the dental implants used nowadays have a

connection which allows a stable more or less rigid connection to an abutment or directly to

3

the dental prosthesis. There are connections allowing the retention via a screw of the

abutment/prosthesis and others in which the abutment is permanently cemented. Connections

usually have various shapes (such as triangles, hexagons, octagons, etc.) and other

mechanism to minimise the risk of movements and screw loosening

The connections more commonly used are the screw-retained ones, since abutments

can be removed if needed. Screw-retained connections can divided in two major groups:

external and internal connections. The external connection is characterised by a mechanism

on the top of the screw to block rotation movements which favours unscrewing. The most

widely used external connection is the "external hexagon" originally used on the Branemark

implant system. The external hexagon connection can be considered the "gold-standard" with

many manufacturers who adopted it, thought there are other types of external connections.

The internal connection is characterised by the presence of the connection mechanism inside

the implant body. There are many different types of internal connections with and without

anti-rotating mechanism and a gold standard here is not easy to identify, though the so called

"conometric" connections have many estimators.

The implant-abutment connection is believed to play an important role in the outcome

of the implant therapy and almost each dental implant manufacturer developed its own

unique connection. These connections are subjected to an aggressive commercial marketing

with many manufactures and clinicians claiming the superiority of one connection over the

others. Interesting to say, despite that osseointegrated dental implants have been in use for

almost half century, not a single well designed and conducted randomised controlled trial

(RCT) has been conducted to specifically investigate the role, if any, of different implant

connection, by evaluating implants having as the only difference their connections. Since

there are no yet any valid evidence-based clinical data evaluating whether one implant

connection could be superior to the others and whether if and how the different connection

4

types could influence the clinical outcome of implant-supported rehabilitations in terms of,

complications, peri-implant marginal bone loss, aesthetics and easy to use, it would be

desirable to have RCTs evaluating these aspects. It would also be interesting to evaluate

whether the preferable connection type could be different depending on the numbers of

implants supporting the same prosthesis (single crown, 2 to 3 connected implants or more

than 3 implants).

The aim of this pragmatic multicentre randomised controlled trial (RCT) of parallel

group design was to evaluate advantages and disadvantages of identical implants with

internal or external connections. The secondary aim was to conduct subgroups analyses

between prostheses supported by one, two to three or more than three implants to see whether

one connection type could be better when different numbers of implants are used. This is the

first report in a series presenting clinical outcome at 1-year post-loading. Ideally, further

reports on this trial will be published after the completion of 5 and 10-year follow-ups. The

present article is reported according to the CONSORT statement for improving the quality of

reports of parallel-group randomised trials (http://www.consort-statement.org/).

MATERIALS AND METHODS

Any patient requiring one implant-supported prosthesis, being 18 years or older, and

able to understand and sign a written informed consent form was eligible for this trial. Only

one prosthesis per patient was to be considered for this trial which could only be supported

by the type of implants dictated by the randomization procedure. This trial was designed as a

pragmatic trial in order to be as close as possible to the clinical reality. Broad inclusion

criteria were used including, for instance, any type of bone quality, any jaw location and

heavy smokers. Clinicians were allowed to choose the treatment option they considered to be

the optimal for the patient to be rehabilitated (for instance flapless implant placement,

5

immediate post-extractive implants, minor augmentation procedures at implant placement,

immediate, early or delayed loading, submerged or non-submerged techniques, etc.) at their

discretion.

Pre-operative radiographs (intra-oral, panoramic, CT scans or other radiographic

examinations at discretion of the operators) together with clinical inspection were used to

determine bone volumes. Exclusion criteria were:

general contraindications to implant surgery

irradiation in the head and neck area

immunosuppressed or immunocompromised patients,

treated or under treatment with intravenous amino-bisphosphonates

untreated periodontitis

poor oral hygiene and motivation

uncontrolled diabetes,

pregnancy or nursing

substance abusers

psychiatric problems or unrealistic expectations

lacking antagonistic occlusal surfaces at the study implant-supported prosthesis at

implant loading

acute/purulent infection in the area intended for implant placement

unrestorable with a retrievable prosthesis to allow individual implant stability assessment

(with exceptions of single implants)

participation in other studies, if the present protocol could not be properly followed

referred only for implant placement

unable to commit to 10-year follow-up.

6

All patients received thorough explanations and signed a written informed consent

form prior to being enrolled in the trial to document that they understood the scope of the

study (including procedures, follow-up evaluations, and any potential risks involved), were

allowed an opportunity to ask questions pertaining to this study, and were apprised of

treatment alternatives. The study was open to qualifying patients without regard to sex or race.

Patients were categorised in three groups according to what they declared: non-

smokers, moderate smokers (up to 10 cigarettes per day), and heavy smokers (more than 10

cigarettes per day). For patients needing more than one implant-supported prosthesis, the

operator could chose which one to include in the study at the screening visit.

Patients were recruited and treated in eight private practices by experienced operators.

Five practices were in Italy (Drs Felice, Grusovin, Gualini, Pistilli and Scarano) and three in

South Korea (Drs Lee 1, Lee 2 and Yoo). Each dentist should have treated 30 patients. All

the follow-up visits were done at the respective treating centres. Originally 10 centres agreed

to participate in the study but two centres did not provide any patient data.

The investigational devices were commercially available tapered titanium screw-

shaped EZ Plus dental implants (MegaGen Implant Co. Lld. Gyeongbuk, South Korea) with

sand-blasted acid-etched surface up to the neck either with external (Fig 1a-d) or internal (Fig

2a-c) connection. The external connection was the standard external hexagon of the

Branemark System (Fig 1a-c), whereas the internal connection was an 11° morse taper

connection (Fig 2a-c) that produces a conical seal forming a cold welding between the

abutment and the implant. The only differences between the two implants apart the

connections are the presence of a bevel at the implant neck of the IC (designed to allow

platform mis-matching) which is not present in the EC design (Figs 1a and 2a) and a different

neck design for the implants with external connection of 3.3 mm diameter (Fig 1d). The neck

was designed differently in order to adapt the standard external hexagon on a small diameter

7

implant. The other difference involved the abutment shape since those designed for the IC

group had to be platform switched (Fig 3 and 4). Operators were free to choose implant

lengths (7, 8.5, 10, 11.5, 13 and 15 mm) and diameters (3.3, 4, 4.5 and 5.5 mm) according to

clinical indications and their preferences.

Clinical procedures

Patients received prophylactic antibiotic therapy: 2 g of amoxicillin (or 600 mg of

clindamycin if allergic to penicillin) one hour prior to surgery and rinsed for one minute with

0.2% chlorhexidine. All patients were treated under local anaesthesia. Tooth extractions,

when needed, were performed as atraumatically as possible attempting to preserve the buccal

alveolar bone. Extraction sockets were carefully cleaned from any remains of granulation

tissue. The decision to elevate or not the flap was left to the individual clinician. The standard

implant site preparation procedure as recommended by the implant manufacturer was used.

In case of soft bone a final drill of one smaller size than the conventional procedure was used

to underprepare the implant site. During implant site preparation bone quality was

subjectively assessed and divided into hard, medium and soft. Once the implant site

preparation was completed, the operator was informed whether the implants to be placed had

to be with external or internal connections, according to a parallel-group study design with

two arms, by opening the sequentially numbered sealed envelope corresponding to the

patient recruitment number. Implants were placed with the neck flush to the crestal bone

level with the exception of post-extractive implants that were placed about two mm below

the palatal bone level and more palatally.

Clinicians were free to decide whether to load the implants immediately early or

conventionally, to submerge or to leave them non-submerged for the healing period they

decided but they had to ensure that both groups were treated in a similar way. Just after

implant placement, intraoral radiographs (baseline) were made with the paralleling technique.

8

If bone levels around the study implants were hidden or difficult to estimate, a second

radiograph was made. Ibuprofen 400 mg was prescribed to be taken 2-4 times a day during

meals, as long as required. Patients were instructed to use 0.2% chlorhexidine mouthwash for

one minute twice a day for two weeks and to avoid brushing and trauma on the surgical sites.

Postoperative antibiotics were only prescribed to patients subjected to bone augmentation

procedures: 1 g of amoxicillin twice a day for six days. Patients allergic to penicillin were

prescribed 300 mg of Clindamycin twice a day for six days. Within one week all patients

were recalled and checked.

Clinicians were also free to choose screw-retained or cemented restorations with

provisional cement, to load the implants directly with definitive restorations, and whether to

use metal-ceramic or metal-composite restorations (single crowns could be also in full

ceramic). Overdentures could also be used.

Periapical radiographs of the study implants were also taken at initial loading and 1

year after loading and individual implants were tested for stability: prostheses connecting

more than one implants were removed and a torque of 20 Ncm was applied to the individual

implants, whereas stability of implant-supported crowns was tested using the handles of two

instruments.

Patients were enrolled in an oral hygiene program with recall visits planned at least

every 6 months for the entire duration of the study.

Outcome measures

This study tested the null hypothesis that there were no differences in the clinical

outcomes between the two connection types against the alternative hypothesis of a

difference. Outcome measures were:

9

Prosthesis failure (primary outcome measure): whether it will not be possible to place

the prosthesis due to implant failures, secondary to implant losses or remake of a

definitive prosthesis for any reasons.

Implant failure (primary outcome measure): implant failure was defined as implant

mobility and/or any infection dictating implant removal or any mechanical failure

rendering the implant unusable, such as implant fracture or deformation of the

implant-abutment connection. The stability of each implant was measured manually

by tightening the abutment screw with a wrench delivering a torque of 20 Ncm or by

assessing the stability of single crowns using the handles of two instruments.

Any complications and adverse events (primary outcome measure) were recorded and

reported by implant types.

Peri-implant marginal bone level changes (secondary outcome measure) evaluated on

intraoral radiographs taken with the paralleling technique at implant placement, at

initial loading and 1 year after loading. Radiographs were scanned in TIFF format

with a 600 dpi resolution, and stored in a personal computer. Peri-implant marginal

bone levels were measured using the UTHSCSA Image Tool 3.0 (The University of

Texas Health Science Center, San Antonio, USA) software. The software was

calibrated for every single image using the known implant neck diameter.

Measurements of the mesial and distal bone crest level adjacent to each implant were

made to the nearest 0.01 mm and averaged at patient level and the at group level. The

measurements were taken parallel to the implant axis. Reference points for the linear

measurements were: the most coronal margin of the implant collar and the most

coronal point of bone-to-implant contact.

Operator preference (secondary outcome measure) for the connection type was

recorded at delivery of the definitive prostheses and 1 year after loading. It was

10

expressed in the following way as: i) internal connection, ii) external connection, iii)

no preference. Reasons for preference were recorded.

At each center a local blind outcome assessor evaluated implant stability. The implant

type was not recognizable when assessing implant stability. One dentist (Hassan Maghaireh)

not involved in the treatment of the patients performed all radiographic assessments without

knowing group allocation, however IC implants could be identified on radiographs due to the

presence of the neck bevel and of platform switched abutments.

Methodological aspects

Prior to this study there were no clinical data from comparative data on which basing

a reliable sample size calculation. It was decided to include 30 patients at each of the 10

planned centres for a total of 300 patients, 150 patients randomised to each group.

Ten computer generated restricted random lists were created. Only one investigator

(Marco Esposito), who was not involved in the selection and treatment of the patients, knew

the random sequence and had access to the random list stored in a pass-word protected

portable computer. The random codes were enclosed in sequentially-numbered, identical,

opaque, sealed envelopes. Only after the implant sites were prepared, the envelope

corresponding to the patient recruitment number was opened and the clinician knew whether

to place an implant with internal or external connection. Therefore, treatment allocations

were concealed to the investigators in charge of enrolling and treating the patients.

All data analyses were carried out according to a pre-established analysis plan. A

dentist with expertise in statistics (Jacopo Buti) analysed the data. Differences in the

proportion of patients with prosthesis failures, implant failures and complications

(dichotomous outcomes) were compared between groups using the Fisher’s exact probability

test. Differences of means at patient level for continuous outcomes (bone levels) between

groups were compared by t-tests. Comparisons between each time points and the baseline

11

measurements were made by paired tests, to detect any changes in marginal peri-implant

bone levels. Differences in radiographic marginal bone levels (continuous outcome variable)

between groups were estimated by creating two linear multilevel models at loading and 1

year changes, respectively - at 2 levels (Centre and Patient). The explicative variables used at

the Patient level were “Implant” (1=Internal Connection, 2=External Connection) and

“Baseline rx bone level” (mm). Differences among centres for dichotomous outcomes were

calculated using the chi-squared test. Pair-wise between-centres differences in mean peri-

implant marginal radiographic bone level changes (continuous outcome) were estimated

using Tukey HSD test after analysis of covariance (ANCOVA) with baseline radiographic

levels as a covariate.

A subgroup analysis, comparing the two implant connections by the following subgroups i)

single crowns; ii) prostheses supported by two to three connected implants; and iii)

prostheses supported by more than three implants, was to be conducted in the attempt to see

whether there was a preferable connection type depending on the number of implants

supporting the same prosthesis. However this was not done since no statistically significant

differences were detected between groups. We my perform this subgroup analysis at 5 year-

post-loading. All statistical comparisons were conducted at the 0.05 level of significance.

RESULTS

Two of the 10 centres never supplied any data. In total at least 331 patients were screened for

eligibility, and 101 patients were not included for the following reasons: 57 patients did not

want to participate into a clinical trial, 26 patients were referred only for implant placement;

seven patients unable to commit to a 10 years follow-up; five patients because were treated

or were under treatment with oral bisphosphonates, four patients had the implant to be

connected to other implant types; two patients for poor oral hygiene/motivation. All patients

12

had their sites treated according to the allocated interventions. Eleven patients with 17

implants dropped-out before the completion of one year post loading (Table 1): with one

centre (Dr Yoo) accounting for 55% of all drop-outs.

Drop-out from the EC group:

one patient with three implants never come back for delivery of the prosthesis due to

lack of time (Dr Lee 1);

one patient with one implant died for throat cancer (Dr Scarano);

one patient with one implant moved out of the country (Dr Yoo).

Drop-out from the IC group:

one patient with one implant died just before the 1-year follow-up without having

experienced implant failures or complications (Dr Lee 1);

one patient with two implants was depressed and did not attend the follow-up but

reported no problem for the implant-supported prosthesis (Dr Grusovin);

one patient with one implant moved to another country (Dr Scarano);

three patients with one, two and three implants, respectively, did not come back for

the one year control (Dr Yoo);

two patients with one implant each did not come back for the prosthetic rehabilitation

(one patient) and for the one year control (one patient) because living too far away

from the clinic (Dr Yoo).

The data of all remaining patients were included in the statistical analyses. The main

protocol deviations are summarised in Table 2. Additional protocol deviations were:

Drs Felice, Scarano and Lee 1 did not supply any radiographic data.

Dr Scarano did not record the number and reasons of those patients screened as

potential candidates for the trial but did not match the inclusion criteria and included

more than one prosthesis per patient in 11 patients.

13

Dr Lee 1 deliberately treated 20 patients instead of 30.

Dr Lee 2 did not record the number and reasons of those patients screened as

potential candidates for the trial but did not match the inclusion criteria and in

addition excluded patients of 'old age'.

Patients were recruited and implants were inserted from January 2009 to November

2011. The follow-up for all patients was 1-year post-loading.

The main baseline patient and intervention characteristics, divided by study group,

are presented in Table 3. There were no apparent significant baseline imbalances between

the two groups. In total 205 EC and 202 IC implants were placed in 117 and 113 patients,

respectively.

Two prostheses failed in the EC group (one was actually delivered with delay since

the implant was found mobile at abutment connection was successfully replaced) versus one

prosthesis (not delivered due to implant failures) in the IC group. There was not statistically

significant difference for patients experiencing prosthesis failures between groups (P=1.0000,

diff.=-0.01, 95% CI -0.05 to 0.03).

Three implants failed in three patients of the EC group versus two implants in one

patient of the IC group (Table 4). There was not statistically significant difference for

patients experiencing implant failures between groups (P=0.6217, diff.=-0.02, 95% CI -0.06

to 0.02). The following implant failures occurred at EC implants:

One patient, smoking up to 10 cigarettes per day, who received one post-

extractive implant (11.5 x 4 mm) in position 21 in soft bone which was

augmented at placement, lost the implant with its provisional crown 7 months

after placement for mobility and inflammation.

14

One non-smoker patient lost one out of two post-extractive implants, in

position 35 (4 x 10 mm) at abutment connection for mobility. It was

successful replaced with another implant.

One non-smoker patient lost one out of four implants, in position 25 (4 x 10

mm) at abutment connection for mobility. The implant was not replaced and a

bar-supported overdenture was made on the three remaining implants.

The following implant failures occurred at EC implants:

One patient, smoking more than 10 cigarettes per day received two implants

(10 x 4 mm and 11.5 x 4 mm) in hard bone in position 46 and 47. The surgery

was painful and pain persisted postoperatively. After 2 weeks, the bone was

exposed and necrotic at both implants, which were removed. These implants

replaced two implants which failed previously and were not replaced.

Nine complications occurred at nine EC implants in nine patients and seven

complications at seven IC implants in seven patients (Table 5). There was not statistically

significant difference for patients experiencing complications between groups (P=0.7964,

diff.= 0.01, 95% CI -0.06 to 0.08). The following complications occurred at EC implants:

The abutment screw become loose once at single implants carrying

provisional crowns in three patients. Crowns were drilled and screw

retightened.

Loosening of the abutment screw under a provisional cross-arch prosthesis.

Screw retightened.

The abutment screw become loose once at a single implant one month after

delivery of a definitive screw-retained crown in one patient. The screw was

retightened at 30 Ncm.

15

The contact point between the implant-supported prosthesis and the adjacent

natural tooth was lost in three patients. Prostheses were unscrewed and

reshaped in the laboratory.

One implant out of two in the same patient, in position 35, was affected by

peri-implantitis at the 1 year follow-up and was surgically treated.

Subsequently the implant failed, however this failure occurred after the 1 year

follow-up and will be reported in a future publication.

IC implants were affected by the following complications:

Post-operative infection with bone exposure leading to failure of two

implants.

Loosening of one healing abutment after one week which was retightened.

The abutment screw become loose once at a single implant carrying a

provisional crown in one patient. The crown was drilled and the screw

retightened.

Loosening of the definitive crown one month after delivery. The abutment

screw was retightened at 35 Ncm.

The abutment screw become loose once on a single implant 4 months after

delivery of a definitive cemented crown in one patient. The screw was

replaced.

Loosening of one definitive crown and one definitive partial prosthesis after

three months in two patients. Screws were retightened at 30 Ncm.

Regarding radiographic peri-implant marginal bone levels, at baseline (implant

placement), there was an almost statistically significant difference (P=0.052; Table 6) of 0.1

mm with higher bone levels at IC implants. Both groups gradually lost marginal peri-implant

bone in a highly statistically significant way at loading and 1 year after loading (P<0.001;

16

Table 7). At loading, patients with EC implants lost an average of 0.59 mm peri-implant

bone versus 0.54 mm for patients with IC implants (Table 7). One year after loading, patients

with EC implants lost an average of 0.98 mm peri-implant bone versus 0.85 mm for patients

with IC implants (Table 7). Multilevel analysis showed no statistically significant difference

between the two groups for peri-implant bone loss at loading (P=0.311; Table 8) and 1 year

after loading (P=0.0629; Table 9).

Operator preference: Five operators (Drs Felice, Pistilli, Yoo, Lee 1 and Gualini) had

no preference between the two connections and three operators preferred the IC connection

(Drs Scarano, Grusovin and Lee 2). Reasons to prefer the IC connection were 'simpler to

handle at abutment connection procedures' (Drs Grusovin and Lee 2) and 'smaller micro-gap

at the implant-abutment connection' (Dr Scarano).

The comparison between centres is presented in Table 10 (clinical outcomes; eight

centres) and 11 (radiographic data; five centres). There were no statistically significant

differences between centres for complications and implant failures, but there are difference

for marginal bone level changes, with Dr Pistilli and Dr Gualini losing about the double

amount of bone than the other three centres.

DISCUSSION

This is the first follow-up report of a planned series aimed at investigating which could be the

advantages and the disadvantages using internal or external connections. In order to perform

a reliable evaluation only the type of connection has to be different, all other implant

characteristics (implant material, surface characteristics and implant shape) remaining

exactly the same. The EZ Plus implant system was chosen because it almost had all the

required characteristics (the main difference were the bevel present at the coronal portions of

IC implants (Fig 2a) and the different neck of the 3.3 mm diameter EC implants (Fig 1d).

17

Another reason for choosing the EZ plus systems was that the implant manufacturer was glad

to sponsor an independent conducted trial to evaluate this aspect. At 1 year post-loading, no

statistically significant differences were observed, the number of complications was low and

similar for both implant types. On one hand, due to the relatively small sample size of this

study and moreover to the short follow-up (only 1 year after loading) it would be hazardous

to conclude that there are no differences between the two connection types. On the other

hands, no statistically or clinically significant differences were observed one year after

loading. In order to draw more reliable conclusions, we need to wait for 5-year data, since the

differences which were not apparent at one year after, may become evident over time. We

decided not to implement the subgroup analysis, planned at protocol preparation, comparing

the outcomes for ì) single crowns; ii) prostheses supported by two to three connected

implants; and ii) prostheses supported by more than three implants, because lack of

statistically significant differences between the two connection type, too few prostheses

supported by more than three implants, and too short follow-up. Such analyses could be still

performed at when the 5-year data will be available.

Some comments could be done on the few complications which might be related

specifically to the connection type. The four early abutment screw loosening (three in EC and

one in IC group) reported by Dr Pistilli at single implants with provisional crowns can be

explained by the habit of the operator to screw manually the abutment screws holding

provisional crowns with torques below 10 Ncm. When placing definitive crowns the surgeon

applied a torque of 25 Ncm and no more screw loosening occurred.

The only parameter which was close to be statistically significant was the

radiographic peri-implant marginal bone levels. Indeed at baseline (implant placement), the

difference was almost statistically significant (P=0.052, Table 6). However, we are

discussing of a difference of 0.1 mm which bears no clinical relevance. One year after

18

loading the statistical model used that considered also the baseline difference showed again a

P-value close to statistical significance (P=0.0629), meaning a bit more than 0.2 mm more

bone loss at EC implants, which again does not appear to be clinical relevant. Only longer

follow-ups could help us to understand whether this specific type of internal connection,

brings additional appreciable benefits over the external connection. We also need to stress

that IC implants were designed to allow platform switching, but in this case, not clinical

advantages could be clearly established by platform switching. Regarding platform

switching, contradictory results have been presented by different authors: some RCTs

showed significantly less bone loss at platform switched implants, whereas other RCTs did

not show any difference. Nevertheless, those RCTs that demonstrated a statistically

significant difference, showed a difference in favour of platform-switched implants of about

0.3 mm at 1 year post-loading which is very close to our findings. The comparisons between

various centres yielded an intriguing observation, while there were no difference for failures

and complication between the eight centres reporting this information, there was statistically

more bone loss (about the double) at two of the five centres who reported radiographic data.

We do not have an tentative explanation or a convincing hypothesis for this difference, but

some difference between treatment protocols could be present.

Five operators had no preference with regard to the internal or external connection

types, whereas three operators preferred IT implants and none ET implants. The numbers

were too low to allow for a statistical analysis, however two dentists pointed out that their

preference was justified by easier prosthetic procedures with the internal connection, which

is understandable. Actually, according to the present findings, operators can choose the

connection type, according to their preferences. It could be also hypothesised that internal

connections are more user-friendly when single implants or prosthesis supported by two or

three implants are used. On the contrary, it may be that in the presence of multiple implants,

19

as when rehabilitating an edentulous jaw, the external connection could be more forgiving at

impression taking than an internal one. However, these are simply hypotheses that need to be

verified.

It is also interesting to observe that despite clinicians were left the option to

choose the time of implant loading, only one and two patients were subjected to immediate or

early loading procedures, respectively. This may suggest that immediately loading

procedures for single implants or small partial fixed prostheses are not so commonly

performed.

There are no other published randomised controlled trials comparing internal versus

external connections, so meaningful comparisons with other similar RCTs cannot be made at

the present stage.

The main limitations of the present trial are: i) the design of the two evaluated

implants was not identical but additional platform switching features were added to the IC

implants which may have slightly favoured IC implants; ii) some of the protocol deviations

and in particular the lack of periapical radiographic from three centres (80 patients in total);

and iii) the high drop-out rate from one of the centres.

Regarding the generalisation of these preliminary results, due to the pragmatic nature

of the present study design, similar results should be obtained by other operators treating

patients with similar procedures.

CONCLUSIONS

It might be imprecise to draw definitive conclusions about the role of internal and external

connections on the clinical outcome due to some minor confounding factors (different neck

design imposing platform switching at IC implants), however, no statistically significant

20

differences were observed, therefore the choice of the type of connection can be simply

based on clinician preference.

ACKNOWLEDGMENTS

Dr Pietro Felice, Antonio Scarano and Jung Sam Lee are acknowledged to have participated

in this trial.

REFERENCES

1. Brånemark P-I, Hansson BO, Adell R, Breine U, Lindström J, Hallén O, et al. Osseointegrated implants in the treatment of the edentulous jaw. Experience from a 10-year period. Stockholm: Almqvist & Wiksell International, 1977.2. Esposito M, Ardebili Y, Worthington HV. Interventions for replacing missing teeth: different types of dental implants. Cochrane Database of Systematic Reviews 2014: Chichester, UK: John Wiley & Sons, Ltd.3. Goldstein H. Multilevel statistical models. London: Edward Arnold, 2003.4. Canullo L, Goglia G, Iurlaro G, Iannello G. Short-term bone level observations associated with platform switching in immediately placed and restored single maxillary implants: a preliminary report. International Journal of Prosthodontics 2009;22:277-282.5. Guerra F, Wagner W, Wiltfang J, Rocha S, Moergel M, Behrens E, et al. Platform switch versus platform match in the posterior mandible - 1-year results of a multicentre randomized clinical trial. Journal of Clinical Periodontology 2014;41:521-529.6. Enkling N, Johren P, Klimberg V, Bayer S, Mericske-Stern R, Jepsen S. Effect of platform switching on peri-implant bone levels: a randomized clinical trial. Clinical Oral Implants Research 2011;22:1185-1192.7. Meloni SM, Jovanovic SA, Lolli FM, Pisano M, De Riu G, De Riu N, et al. Platform switching vs regular platform implants: nine-month post-loading results from a randomised controlled trial. European Journal of Oral Implantology 2014;7:257-265.

21

Table 1. Summary of drop-outs by centre, in parenthesis the reasons for dropping-out

(N = number of patients)

Felice(N=30)

Pistilli(N=30)

Scarano(N=30)

Grusovin(N=30)

Yoo(N=30)

Lee1(N=20)

Gualini (N=30)

Lee2(N=30)

Total (N=230)

EC (N=117) 0 0 1 (death) 0 1 (moved) 1 (death) 0 0 3

IC (N=113) 0 0 1 (moved) 1 (depression) 5 (did not come back) 1 (no time) 0 0 8

Table 2. Summary of protocol deviations by centre (N = number of patients)

Felice(N=30)

Pistilli(N=30)

Scarano(N=30)

Grusovin(N=30)

Yoo(N=30)

Lee1(N=20)

Gualini(N=30)

Lee2(N=30)

Total(N=230)

EC (N=117)

0 0 0 2** 0 0 2*****

1******

3*

2**

10

IC (N=113)

0 2*** 0 1* 1**** 0 1****

1*****

1* 7

*Prostheses attached to other implant types having the same connections.

**Used Ex-Feel implants with correct connection instead.

***Took panoramic and not periapical radiographs at 1 year after loading.

****Missing periapical radiograph at loading.

*****Missing periapical radiograph at 1 year after loading.

******Took unreadable panoramic instead of periapical radiograph at 1 year.

22

Table 3. Patient and intervention characteristics.

EC [n=116] (%) IC [n=114] (%)Females (%) 68 (59%) 62 (54%)Mean age at implant insertion (range) 51.2 ± 14.5 (22-79) 52.4 ± 14 (20-80)Smoking up to 10 cigarettes/day (%) 21 (18%) 18 (16%)Smoking more than 10 cigarettes/day (%) 13 (11%) 19 (17%)Implants in upper jaws (%) 84 (41%) 79 (39%)Implants in lower jaws (%) 121 (59%) 123 (61%)Implants in incisor position 14 17Implants in canine position 7 15Implants in premolar position 80 66Implants in molar position 103 104Implants in hard bone 48 46Implants in medium bone 121 131Implants in soft bone 36 25Implants with 3.3 mm diameter 23 17Implants with 4 mm diameter 94 115Implants with 4.5 mm diameter 1 0Implants with 5.5 mm diameter 87 71Implants 7 mm long 0 1Implants 8.5 mm long 36 25Implants 10 mm long 50 63Implants 11.5 mm long 68 63Implants 13 mm long 51 50Post-extractive implants (%) 41 (20%) 30 (15%)Implants in augmented sites (%)* 63 (31%) 62 (31%)Implants inserted flapless (%) 99 (48%) 101 (50%)Patients with implants submerged (%) 85 (73%) 94 (82%)Single crowns (%) 59 (51%) 71 (62%)Partial fixed prostheses (%) 50 (43%) 39 (34%)Cross-arch fixed prostheses (%) 3 (3%) 1 (1%)Overdentures (%) 3 (3%) 2 (2%)Patients with implants immediately loaded (within 1 week) (%)

0 (0%) 1 (1%)

Patients with implants early loaded (between 1 week and 2 months) (%)

2 (2%) 0 (0%)

Patients with implants conventionally loaded (after 3 months) (%)

114 (98%) 112 (98%)

*Including augmented post-extractive sites at implant placement

23

Table 4: Summary and comparisons expressed at implant level (N= number of

implants) of implant failures up to 1-year after loading by groups.

Felice(N=54)

Pistilli(N=57)

Scarano(N=80)

Grusovin(N=41)

Yoo(N=46)

Lee1(N=24)

Gualini(N=54)

Lee2(N=51)

Total(N=407)

EC(N=205)

2 (N=33) 0 (N=27) 0 (N=32) 0 (N=22) 0 (N=28) 1 (N=18) 0 (N=27) 0 (N=20) 3 (N=205)

IC(N=202)

0 (N=21) 0 (N=30) 0 (N=48) 0 (N=19) 0 (N=18) 0 (N=8) 2 (N=27) 0 (N=31) 2 (N=202)

p-valuea 0.5157 1 1 1 1 1 0.4916 1 1aFisher Exact test; *statistically significant differences.

Table 5: Summary and comparisons expressed at implant level (N= number of

implants) of complications up to 1-year after loading by groups.

Felice(N=54)

PistilliN=57)

Scarano (N=80)

Grusovin(N=41)

Yoo (N=46)

Lee1(N=24)

Gualini(N=54)

Lee2(N=51)

Total(N=407)

EC (N=205)

0 (N=33)

3 (N=27) 0 (N=32) 1 (N=22) 1 (N=28) 0 (N=18) 1 (N=27) 3 (N=20) 9 (N=205

IC (N=202)

0 (N=21)

1 (N=30) 1 (N=48) 3 (N=19) 1 (N=18) 0 (N=8) 1 (N=27) 0 (N=31) 7 (N=202)

p-valuea 1 0.3360 1 0.3210 1 1 1 0.0574 0.7997aFisher Exact test; *statistically significant differences.

24

Table 6: Mean radiographic peri-implant marginal bone levels between groups and time

periods.

Implant placement Loading* 1 year after loading*

N Mean (SD) 95% CI N Mean (SD) 95% CI N Mean (SD) 95% CI

EC implants 75 0.18 (0.42) [0.09;0.28] 75 0.77 (0.62) [0.62;0.91] 71 1.17 (0.92) [0.95;1.39]

IC implants 73 0.07 (0.22) [0.02;0.12] 71 0.60 (0.70) [0.44;0.77] 64 0.91 (0.75) [0.73;1.10]

Difference

(SE) 95% CI

p-value

73 0.11 (0.05) [-0.01;0.21]

0.0502

*All changes from baseline (paired t-test) statistically different (P<0.001). SD = Standard

deviation; CI = Confidence interval.

Table 7: Descriptive statistics of mean changes in peri-implant marginal bone levels at

different time periods between groups.

Baseline – loading* Baseline – 1 year after loading*

N Mean (SE) 95% CI N Mean (SE) 95% CI

EC implants 75 0.59 (0.07) [0.44;0.73] 71 0.98 (0.12) [0.74; 1.22]

IC implants 70 0.54 (0.08) [0.38;0.69] 63 0.85 (0.09) [0.68; 1.03]

* Mean differences within group from paired t-test analysis; SD = Standard deviation; CI =

Confidence interval.

25

Table 8: Multilevel model for the radiographic peri-implant marginal bone levels changes

between groups at loading.*Statistically significant difference; SE = Standard error; CI = Confidence interval.

Radiographic peri-implant marginal bone levels changes at loading

Term Estimate SE P-value [95% CI]

Intercept 0.47 0.18

Patient level

Baseline rx bone level -0.49 0.16 0.0017* [-0.79.;-0.18]

Implant (1=Int; 2=Ext) 0.10 0.10 0.3111 [-0.10;0.30]

Variances

u2 0.03 0.03

e 2 0.37 0.04

Theoretic model:

Loading-Bas rx = β0ij + β1ij Implant + β2ij Baseline rx bone level

u2 and e

2 indicate the variances at the Centre and at the Patient level, respectively. In the theoretic model

formula, the subscript j refers to the Centre level. The subscript i refers to the Patient level. β0ij is the 'intercept'.

SE is the standard error.

Data not provided for centres Felice, Scarano and Lee1.

26

Table 9: Multilevel model for the radiographic peri-implant marginal bone levels changes

between groups at 1 year.*Statistically significant difference; SE = Standard error; CI = Confidence interval.

Radiographic peri-implant marginal bone levels changes at 1 year

Term Estimate SE P-value [95% CI]

Intercept 0.67 0.27

Patient level

Baseline rx bone level -0.82 0.19 <0.0001* [-1.20;-0.44]

Implant (1=Int; 2=Ext) 0.24 0.13 0.0629 [-0.01;0.50]

Variances

u2 0.17 0.11

e 2 0.55 0.07

Theoretic model:

1 year-Bas rx = β0ij + β1ij Implant + β2ij Baseline rx bone level

u2 and e

2 indicate the variances at the Centre and at the Patient level, respectively. In the theoretic model

formula, the subscript j refers to the Centre level. The subscript i refers to the Patient level. β0ij is the 'intercept'.

SE is the standard error.

Data not provided for centres Felice, Scarano and Lee1.

Table 10. Comparisons between study centres for the various outcome measures expressed at patient level (N= number of patients).

Felice (N=30)

Pistilli (N=30)

Scarano (N=30)

Grusovin (N=30)

Yoo (N=30)

Lee1 (N=20)

Gualini (N=30)

Lee2 (N=30)

P-value

Prosthesis failures(N=230)

1 0 0 0 0 1 1 0 0.693a

Implant failures(N=230)

2 0 0 0 0 1 1 0 0.366 a

Complications (N=230)

0 4 1 4 2 0 2 3 0.213 a

aChi-Square test

Table 11. One year radiographic peri-implant marginal bone levels changes from baseline

and pair-wise differences between centres (N = number of patients).

27

Mean (SE) 95%CI p-value Between centres

Differencesa

Pistilli (N = 27) 1.17 (0.14) [0.88;1.46] <0.0001* A

Grusovin (N = 28) 0.64 (0.11) [0.42;0.87] <0.0001* B

Yoo (N = 24) 0.66 (0.16) [0.33;0.99] 0.0004* B

Gualini (N = 25) 1.55 (0.24) [1.06;2.04] <0.0001* A

Lee2 (N = 30) 0.63 (0.12) [0.39;0.87] <0.0001* B

Estimates are expressed as means (SE). *Within-centre statistically significant differences from baseline. Centres not connected by the same letter are significantly different; aanalysis of Covariance (ANCOVA) with baseline radiographic peri-implant bone level as a covariate.

28

Figures

Fig 1a

Fig 1b

29

Fig 1c

Fig 1d

Fig 1a-d: EZ Plus implant with external connection: a); sagittal view; b) occlusal view of the

external connection; c) sagittal section showing the external connection; d) the 3.3

mm diameter has a different neck design to have the same connection it is represented

here with the implant mount.

30

Fig 2a

Fig 2b

Fig 2c

Fig 2a-c: EZ Plus implant with internal connection: a); sagittal view; b) occlusal view of the

internal connection; c) sagittal section showing the internal connection.

31

Fig 3: 1-year post-loading periapical radiograph of one Ez Plus implant with external

connection included in this study (courtesy of Dr Sang Taek Lee). Implant originally placed

with the neck level to the bone. Some marginal bone loss occurred over time.

Fig 4: 1-year post-loading periapical radiograph of 2 Ez Plus implants with internal

connection included in this study. Please note that IC implants had to be platform-switched

due to the implant design (courtesy of Dr Sang Taek Lee). Implants originally placed with

the neck level to the bone. Marginal bone levels remained virtually stable.

32