2262 diabetes care volume 42, december 2019

Efficacy Safety and Tolerability ofOral Semaglutide Versus PlaceboAdded to Insulin With or WithoutMetformin in Patients With Type2 Diabetes The PIONEER 8 TrialDiabetes Care 2019422262ndash2271 | httpsdoiorg102337dc19-0898

OBJECTIVE

To investigate the efficacy safety and tolerability of oral semaglutide added toinsulin with or without metformin

RESEARCH DESIGN AND METHODS

Patients with type 2 diabetes uncontrolled on insulin with or without metforminwere randomized to oral semaglutide 3mg (N5 184) 7mg (N5 182) or 14mg (N5

181) or to placebo (N5 184) in a 52-week double-blind trial End points were changefrom baseline to week 26 in HbA1c (primary) and body weight (confirmatorysecondary) Two estimands were defined treatment policy (effect regardless of trialproduct discontinuation or rescue medication) and trial product (effect assumingtrial product continuation without rescue medication) in randomized patients

RESULTS

Oral semaglutide was superior to placebo in reducing HbA1c (estimated treatmentdifference [ETD] ndash05 [95CI ndash07 ndash03] ndash09 [ndash11 ndash07] and ndash12 [ndash14 ndash10]for 3 7 and 14 mg respectively P lt 0001) and body weight (ETD 209 kg [95 CI218200]220 kg [230210] and233 kg [242223] P5 00392 for 3mg Ppound00001 for 7 and 14 mg) at week 26 (treatment policy estimand) Significantly greaterdose-dependent HbA1c and body weight reductions versus placebo were achievedwith oral semaglutide at weeks 26 and 52 (both estimands) The most fre-quent adverse event with oral semaglutide was nausea (114ndash232 of patientsvs 71 with placebo mostly mild to moderate)

CONCLUSIONS

Oral semaglutide was superior to placebo in reducing HbA1c and bodyweight whenadded to insulin with or without metformin in patients with type 2 diabetes Thesafety profile was consistent with other glucagon-like peptide 1 receptor agonists

Glucagon-like peptide 1 receptor agonists (GLP-1RAs) reduce HbA1c with a low risk ofhypoglycemia and favorable effects on body weight (12) Furthermore some GLP-1RAs provide cardiovascular benefits and are recommended by diabetes andcardiology guidelines for patients with concomitant cardiovascular disease (34)Combined with insulin GLP-1RAs reduce HbA1c and body weight from baselinewithout increasing hypoglycemia (5ndash9) Semaglutide is a GLP-1RA and its

1Lunenfeld-Tanenbaum Research Institute MountSinai Hospital University of Toronto TorontoOntario Canada2Division of Endocrinology Diabetes andHypertension Brigham and Womenrsquos HospitalHarvard Medical School Boston MA3MedStar Health Research Institute HyattsvilleMD4Division of Endocrinology and MetabolismUniversity of North Carolina School of MedicineChapel Hill NC5Department of Endocrinology LrsquoInstitut duThorax CIC INSERM 1413 CHU Nantes Uni-versite de Nantes Nantes France6Department of Family Medicine Schulich Schoolof Medicine and Dentistry Western UniversityLondon Ontario Canada7Novo Nordisk AS Soslashborg Denmark8Department of Metabolic Medicine Faculty ofLife Sciences Kumamoto University KumamotoJapan

Corresponding author Bernard Zinman zinmanlunenfeldca

Received3May2019andaccepted10September2019

Clinical trial reg no NCT03021187 clinicaltrialsgov

This article contains Supplementary Data onlineat httpcarediabetesjournalsorglookupsuppldoi102337dc19-0898-DC1

This article is featured in a podcast available athttpwwwdiabetesjournalsorgcontentdiabetes-core-update-podcasts

A complete list of investigators in the PIONEER8trial is provided in the Supplementary Data

copy 2019 by the American Diabetes AssociationReaders may use this article as long as the workis properly cited the use is educational and notfor profit and the work is not altered More infor-mation is available at httpwwwdiabetesjournalsorgcontentlicense

Bernard Zinman1 Vanita R Aroda23

John B Buse4 Bertrand Cariou5

Stewart B Harris6 Soslashren Tetens Hoff7

Karen Boje Pedersen7

Mads Jeppe Tarp-Johansen7 and

Eiichi Araki8 for the

PIONEER 8 Investigators

2262 Diabetes Care Volume 42 December 2019

EMER

GINGTH

ERAPIES

DRUGSANDREG

IMEN

S

subcutaneous once-weekly formulationimproved glycemic control and reducedbody weight from baseline when usedalongside basal insulin in patients withtype 2 diabetes (6)An oral formulation of semaglutide

has been developed and is the first oralGLP-1RA to enter phase 3 trials Aspeptides have low oral bioavailabilityoral semaglutide is coformulated withthe absorption enhancer sodiumN-(8-[2-hydroxybenzoyl] amino) caprylatewhichfacilitates semaglutide absorption acrossthe gastric mucosa (10)This article reports the findings of the

Peptide Innovation for Early DiabetesTreatment 8 (PIONEER 8) trial whichinvestigated the efficacy safety and tol-erability of oral semaglutide added ontoinsulin (basal basal-bolus or premixed)with or without metformin in patientswith uncontrolled type 2 diabetes


Trial DesignThis randomized double-blind placebo-controlled parallel-group trial was con-ducted at 111 sites in nine countries(Supplementary Appendix 1) between2 February 2017 and 18 January 2018(NCT03021187) There was a 2-weekscreening period 52-week treatmentperiod and 5-week follow-up period(Supplementary Fig 1A)Patients were randomized 1111 to

once-daily oral semaglutide 3 7 or 14mgor placebo using an interactive web re-sponse system (Supplementary Appendix2) Randomization was stratified by pa-tientsrsquo country of origin (Japanese ornon-Japanese) and background treatment(metformin or no metformin basal basal-bolus or premixed insulin)A 20 reduction in total daily insu-

lin dosage was recommended at ran-domization and maintained to week 8(Supplementary Fig 1B) unless an in-crease was required to prevent acutemetabolic deterioration The treatmentperiod was then split into two stagesdefined by restrictions in total daily in-sulin dosage It could be altered duringweeks 8ndash26 without exceeding the pre-randomization dosage and was freelyadjustable at the investigatorrsquos discretionduring weeks 26ndash52 Throughout thetrial the total daily insulin dosage could bereduced as needed Itwas recommendedthat adjustments were made based onthe lowest of three self-measured blood

glucose (SMBG) values preferably mea-sured on 3 consecutive days prior toeach phone contactsite visit with theaim of obtaining a fasting plasma glucose(FPG) concentration of 40ndash55 mmolL(71ndash99 mgdL) and HbA1c 70 (53mmolmol) (Supplementary Appendix3) In brief dosage was increased inincrements of 2unitsbasedonFPGvaluesstarting at 2 units for 56ndash70 mmolL(100ndash126 mgdL) up to 8 units if 90mmolL (162 mgdL) For patients onbasal-bolus insulin takenmore than oncedaily it was recommended to titrateeach dose separately For patients on basal-bolus insulin recommendations wereprovided only for the basal component

The trial protocol was approved by theinstitutional review boardindependentethics committee at each site and thetrial was conducted in accordance withInternational Council on Harmonization(ICH) Good Clinical Practice guidelinesand the Declaration of Helsinki Allpatients provided written informedconsent

Two different scientific questions re-lated to the efficacy objectives were ad-dressed through the definition of twoestimands the treatment policy estimandand the trial product estimand Bothestimands were defined based on inter-actions with regulatory agencies (11)

The treatment policy estimand ad-dressed the question of the treatmenteffect for all randomized patients regard-less of trial product discontinuation or useof rescue medication This estimand re-flects the intention-to-treat principle asdefined in ICH E9 (12) This estimandreflects the effect of initiating treatmentwith oral semaglutide compared withinitiating treatment with placebo bothpotentially followed by either discontinu-ation of trial product andor addition of orswitch to another glucose-lowering drug

The trial product estimand addressedthe question of the treatment effect forall randomized patients under the as-sumption that all patients remained ontrial product for the entire planned du-ration of the trial and did not use rescuemedication This estimand aims at re-flecting the effect of oral semaglutidecompared with placebo without the con-founding effect of rescue medicationThe statistical analysis that was appliedto estimate this estimand is similar tothat used in the majority of previouslypublished phase 3a diabetes trials (13)

Trial product discontinuation and initi-ation of rescue medication were ac-counted for by the treatment policystrategy for the treatment policy estimandand by the hypothetical strategy for thetrial product estimand as defined in draftICH E9 (R1) (14) Further details are pro-vided in Supplementary Appendix 4

Patient PopulationAdult patients with type 2 diabetesdiagnosed $90 days before screeningwith baseline HbA1c 70ndash95 (53ndash80mmolmol) were enrolled Patients wererequired to be on a stable regimen ofbasal basal-bolus (in any combination)or premixed insulin (including combina-tions of soluble insulin) at$10 unitsdayfor $90 days before screening If usedconcomitant metformin was required tobe at a stable dosage ($1500mg daily orthe maximum tolerated dosage) for$90 days before screening Asidefrom metformin the insulin regimensdescribed above or short-term (14 days)changes in insulin dosage for acuteillness use of any other glucose-loweringmedication was not allowed in the 90days before screening Full eligibilitycriteria are provided in SupplementaryTable 1

Drug AdministrationAs the presence of food or liquid in thestomach impairs absorption of oralsemaglutide (10) patients were instructedto administer trial product in the morningin a fasting state with120mL (4 fl oz)water then towait at least 30min beforethe first meal of the day or taking otheroral medication Tablets were not tobe broken or chewed These instructionswere to ensure sufficient absorption oforal semaglutide Patients randomized tooral semaglutide 3mgwere initiated andremained on the 3-mg dose Those ran-domized to 7 and14mgbegan treatmentat 3 mg and the dose was escalated to7 mg after 4 weeks and to 14 mg after afurther 4 weeks until the randomizeddose was achieved Patients and inves-tigators were blinded to all dose esca-lation steps

Glucose-lowering rescue medication(either newglucose-loweringmedicationor intensification of existing medication)was available to patients taking trialproduct who had persistent or unac-ceptable hyperglycemia based on pre-defined FPG and HbA1c rescue criteria

carediabetesjournalsorg Zinman and Associates 2263

(two measures of FPG 111 mmolL[200 mgdL] from week 16 onwardandor HbA1c 85 [694 mmolmol]from week 26 onward) Intensificationwas defined as a 20 increase inthe dose of existing medication frombaseline maintained for either two ormore visits (for insulin) or$21 days (forother medications) Upon trial productdiscontinuation (either prematurely or atthe end of the treatment period at week52) patients had their total daily insulindosage adjusted andor switched to asuitable marketed product at the inves-tigatorsrsquo discretion The use of GLP-1RAswas prohibited until after the follow-upvisit 5 weeks after the last dose of trialproduct Patients continued in the trialafter receiving rescue medication or pre-maturely discontinuing trial product

Study End Points and AssessmentsThe primary end point was change inHbA1c and the confirmatory secondaryend point was change in body weightboth from baseline to week 26Supportive secondary end points as-

sessed at weeks 26 and 52 were changesfrom baseline in HbA1c (week 52 only)body weight (week 52 only) total dailyinsulindosage FPGSMBG7-pointprofile(mean and mean postprandial incre-ment) BMI waist circumference andfasting lipid profile whether patientsachieved HbA1c 70 (53 mmolmol)and65 (48 mmolmol) body weightloss$5 and$10 and composites ofHbA1c 70 (53 mmolmol) withouthypoglycemia (treatment-emergent se-vere [defined according to the AmericanDiabetesAssociation (ADA) classification]or blood glucosendashconfirmed [31mmolL(56 mgdL)]) symptomatic hypoglycemiaand without body weight gain andHbA1c reduction $10 (109 mmolmol)and body weight loss$3 Changes frombaseline to weeks 26 and 52 in the fol-lowing patient-reported outcomeswere also assessed Short Form (SF)36v2 Health Survey (Acute Version) Im-pact of Weight on Quality of Life-LiteClinical Trial Version (IWQOL-Lite-CT) and the Diabetes Treatment Sat-isfaction Questionnaires (DTSQs)Safety end points included the number

of treatment-emergent adverse events(AEs) during exposure to trial productthe number of severe or blood glucosendashconfirmed symptomatic hypoglycemic ep-isodes (as defined above for the composite

end point) and whether a patient experi-enced such episodes and changes frombaseline in laboratory assessments andvital signs

Statistical AnalysisA sample size of 180 patients per treat-ment arm was calculated to provide90 power to jointly confirm HbA1csuperiority of oral semaglutide over pla-cebo at all dose levels at week 26

Efficacy analyses were based on allrandomized patients The confirmationof efficacy of oral semaglutide on changein HbA1c and body weight both frombaseline to week 26 was based on aweighted Bonferroni closed-testing strat-egy (15) (outlined in SupplementaryAppendix 5) to control the overall typeI error for the hypotheses evaluated bythe treatment policy estimand

The treatment policy estimand wasestimated by a pattern mixture modelusing multiple imputation to handlemissing week 26 data for both confirma-tory end points Data collected at week26 from all randomized patients irrespec-tive of premature discontinuation of trialproduct or initiation of rescue medica-tion were included in the statistical anal-ysis Imputation was done within groupsdefined by trial product and treatmentstatus at week 26 Both the imputationand the analysis were based on ANCOVAmodels The results were combined byuse of Rubinrsquos rule (16)

The trial product estimand was esti-mated by a mixed model for repeatedmeasurements that used data collectedprior to premature trial product discon-tinuation or initiation of rescue medica-tion from all randomized patients

Safety end points were assessed usingthe safety analysis set (all randomizedpatients exposed to one or more dosesof trial product) and evaluated both ontreatment (iewhile receiving trial prod-uct regardless of rescue medication use)and in trial (ie while in the trial regard-less of trial product discontinuation orrescue medication use)

All statistical analyses were performedusing SAS version 94M2 Further detailscan be found in Supplementary Appendix 5

RESULTS

PatientsOf the 1038 patients screened 731were randomized (oral semaglutide 3mg N 5 184 7 mg N 5 182 14 mg

N5181placeboN5184)and included inthe efficacy analyses (Supplementary Fig2) All patients except one (oral semaglu-tide 7 mg) were exposed to trial productand included in the safety analysis set

Demographics and baseline disease char-acteristics are presented in SupplementaryTable 2Overall 395 (540) patientsweremale 376 (514)werewhite 263 (360)were Asian and 49 (67) were black orAfrican American Mean age was 61 yearsmean HbA1c was 82 (66 mmolmol)mean body weight was 859 kg and meandiabetes duration was 150 years

Overall 697 (953) patients com-pleted the trial and trial product wasdiscontinued prematurely by 24 (130)34 (187) 37 (204) and 22 (120)patients for oral semaglutide 3 7 and14 mg and placebo respectively By week26 5 (27) 2 (11) 4 (22) and9 (49) patients and by week 5254 (293) 33 (181) 31 (171)and 67 (364) patients had initiatedrescue medication for oral semaglutide 37 and 14 mg and placebo respectively(Supplementary Table 3) The increaseduse of rescue medication fromweek 26 to52 (in most cases a 20 increase intotal daily insulin dosage) reflects thatinsulin was freely adjustable duringweeks 26ndash52 to reach an HbA1c 70

Background metformin was used by491 (672) patients The total numberof patients on each insulin regimen atscreening was 306 (419) 284 (389)and 129 (176) for basal basal-bolusand premixed respectively (SupplementaryTable 2) Twelve patients were recordedto be on insulin regimens not defined inthe protocol five were on regimens con-sidered equivalent to those in the protocoland continued in the trial and seven wererandomized in error Of these seven pa-tients one was never exposed to trialproduct and treatment was discontinuedfor the remaining six upon discovery Forclinical reporting these patients wereassigned to the insulin regimen as orig-inally assessed by the investigator

At baseline the overall mean (SD) totaldaily insulin dosagewas 58 units (57 units)The mean total daily insulin dosage atbaseline was slightly greater in the oralsemaglutide 3- and 7-mg arms than inthe 14-mg and placebo arms (61 and63 units vs 53 and 55 units respectively)(Supplementary Table 2) A 20 reductionin total daily insulin dosage was recom-mended when initiating trial product The

2264 Oral Semaglutide in Patients Taking Insulin Diabetes Care Volume 42 December 2019

majority ofpatients (753 [n5546]) hadtheir insulin dosage reduced by 15ndash25For the remaining patients the total dailyinsulin dosage was reduced by15 for84 (n5 61) by25 for 34 (n5 25)and was unchanged for 124 (n 5 90)There was no clear association betweenHbA1c at screening and initial insulindosage reduction

Glycemic ControlFor the treatment policy estimand theestimated mean changes from base-line in HbA1c at week 26 were ndash06(ndash6 mmolmol) ndash09 (ndash10 mmolmol)ndash13 (ndash14 mmolmol) and ndash01(ndash1 mmolmol) for oral semaglutide3 7 and 14 mg and placebo re-spectively Compared with placeboHbA1c reductions were superior forall doses of oral semaglutide with esti-mated treatment differences (ETDs) ofndash05 (95 CI ndash07 ndash03) (ndash5 mmolmol[ndash8 ndash3] P 00001) ndash09 (ndash11 ndash07)(ndash10 mmolmol [ndash12 ndash7] P 00001)and ndash12 (ndash14 ndash10) (ndash13 mmolmol[ndash15 ndash11] P 00001) for the 3- 7-and 14-mg doses respectively (Fig 1)Sensitivity analyses were consistent withthese findings (Supplementary Table 4and Supplementary Fig 3) In additionthere were statistically significantly greaterHbA1c reductions from baseline for all oralsemaglutide doses versus placebo at week26 for the trial product estimand and atweek 52 for both estimands (Fig 1) Fur-thermore the observed proportions ofpatients achieving HbA1c 70 (53mmolmol) and 65 (48 mmolmol)were greater with oral semaglutide com-pared with placebo The odds of achiev-ing these targets were statisticallysignificantly greater with oral sema-glutide than with placebo (both esti-mands) (Table 1 and SupplementaryTable 5)At week 26 the total daily insulin

dosage was reduced from baseline inall treatment arms (Fig 1) By week 52total daily insulin was reduced from base-line with oral semaglutide 7 and 14 mgand increasedwithoral semaglutide3mgand placebo These changes from base-line were statistically significantly differ-ent with oral semaglutide versus placeboat weeks 26 and 52 except for the 3-mgdose at week 26 for the treatment pol-icy estimand As previously mentionedmany patients (mainly those on oral sema-glutide 3 mg and placebo) increased their

total daily insulin dosage from baselineby 20 during the freely adjustableinsulin treatment period (weeks 26ndash52)(Supplementary Table 3) As this wasconsideredrescuemedication thisaffectedthe results for this end point for the trialproduct estimand where only data priorto initiation of rescue medication wereused when estimating the results (Fig 1)

Changes from baseline in FPG (Fig 1)and 7-point SMBG means (Table 1) werestatistically significantly greater with oralsemaglutide than placebo at weeks26 (except for FPG with 3 mg for thetreatment policy estimand) and 52 forboth estimands

Body WeightFor the treatment policy estimand theestimated mean body weight changesfrom baseline at week 26 were ndash14 ndash24ndash37 and ndash04 kg for oral semaglutide 37 and 14 mg and placebo respectivelyCompared with placebo these bodyweight reductions were superior for alldoses of oral semaglutide with an ETDof ndash09 kg (95CI ndash18 ndash00P500392)ndash20 kg (ndash30 ndash10 P 5 00001) andndash33 kg (ndash42 ndash23 P 00001) for the3- 7- and 14-mg doses respectively(Fig 2) Sensitivity analyseswereconsistentwith these findings (Supplementary Table 4and Supplementary Fig 3) There werestatistically significantly greater reductionsin body weight from baseline with all oralsemaglutide doses compared with placebofor thetrialproductestimandatweek26andfor both estimands at week 52 (Fig 2 andSupplementary Table 5) Furthermore theobserved proportions of patients achiev-ing body weight loss $5 were greaterwith oral semaglutide than with placeboThe odds of achieving this outcome werestatistically significantly greater with oralsemaglutide than with placebo (bothestimands) (Fig 2)

All oral semaglutidedoses reducedBMIstatistically significantly versus placeboat weeks 26 and 52 (both estimands)(SupplementaryTable5)Results forotherbody weightndashrelated end points are pre-sented in Supplementary Table 5

Other OutcomesThe observed proportions of patientsachieving HbA1c 70 (53 mmolmol)without hypoglycemia and without bodyweight gain were greater and the oddsof achieving the outcome statisticallysignificantly greaterwithoral semaglutide

compared with placebo (both esti-mands) (Table 1) Oral semaglutide treat-ment tended to improve the fasting lipidprofile from baseline Reductions in totalcholesterolwere statistically significantlygreater with all oral semaglutide dosescompared with placebo for both esti-mands at weeks 26 and 52 except forthe 3-mg dose at week 52 for the trialproduct estimand (Supplementary Table5) Results for the patient-reported out-comes are presented in SupplementaryResults 1 and Supplementary Figs 4ndash6and for the other supportive secondaryend points in Supplementary Table 5

SafetyComparable proportions of patients ex-perienced at least one AE while on treat-ment (Table 2) Gastrointestinal disordersoccurred most frequently in the oralsemaglutide 7- and 14-mg arms (3 mg391 [n 5 72] 7 mg 448 [n 5 81]14 mg 503 [n 5 91]) whereas infec-tions and infestations were most commonin the oral semaglutide 3-mg (397 [n573]) and placebo (435 [n5 80]) armsThe most frequently reported AEs werenausea with oral semaglutide (dose-dependently affecting 114ndash232 [n 521ndash42] of patients) and nasopharyngitiswith placebo (147 [n5 27] of patients)(Table 2) Of the nausea events themajority were of mild or moderate severityandofshortduration(SupplementaryFig7)

Serious AEs were reported by 136(n5 25) 105 (n5 19) 66 (n5 12)and 92 (n5 17) of patients in the oralsemaglutide 3- 7- and 14-mg and pla-cebo arms respectively (Table 2) Trialproduct was prematurely discontinuedbecause of AEs by 71 (n 5 13) 88(n516) 133 (n524) and27 (n55)of patients for oral semaglutide 3 7and 14 mg and placebo respectivelywith gastrointestinal disorders being themost frequent cause (SupplementaryTable 6)

There was one pregnancy during thetrial in a patient exposed to trial product(oral semaglutide 7 mg) treatment wasdiscontinued and the patient elected tohave a termination

Very few patients experienced severehypoglycemic episodes (Table 2) Theproportions of patients with a severeor blood glucosendashconfirmed symptom-atic hypoglycemic episode were similarbetween patients receiving oral semaglu-tide and placebo (3 mg 283 [n 5 52]


Figure 1mdashGlycemic controlndashrelated efficacy end points A Observed absolute HbA1c over time placebo oral semaglutide 3 mg oralsemaglutide 7 mg oral semaglutide 14 mg B Estimated changes from baseline in HbA1c C Estimated changes from baseline in FPG D Estimatedchanges from baseline in total daily insulin dosage Treatment policy estimand ANCOVA for continuous end points and logistic regression for binary endpoints using data irrespective of discontinuation of trial product or initiation of rescue medication Missing values were imputed by a pattern mixture modelusing multiple imputation Pattern was defined by randomized trial product and treatment status Trial product estimand mixed model for repeatedmeasurements for continuous end points and logistic regression for binary end points Data collected after discontinuation of trial product or initiationof rescue medication were excluded For binary end points missing values were imputed from patients randomized to the same trial product usingsequential multiple imputation Statistically significant ETD versus placebo in favor of oral semaglutide P values are unadjusted two-sided P valuesfor the test of no difference


7 mg 260 [n 5 47] 14 mg 265[n5 48] placebo 293 [n5 54]) (Table2) Across all treatment arms the greatestnumber of hypoglycemic episodes oc-curred in patients on basal-bolus insulinComparable proportions of patients

experienced diabetic retinopathyndashrelated AEs (Supplementary Table 7) allof which were mild or moderate inseverity Retinopathy events were iden-tified during routine examination for40 patients (10 per treatment arm)and 8 patients required treatmentThe prevalence of external event adju-dication committee (EAC)ndashconfirmedcardiovascular events and acute kidneyinjury events during the trial was lowand similar across treatment arms(Supplementary Table 8) Few patientshad EAC-confirmed malignant neoplasmsand there were no EAC-confirmed eventsof acute pancreatitisThere were three deaths during the trial

all of which occurred on treatment with

oral semaglutide 14 mg (SupplementaryTable 8) Of these patients none reportedsevere or blood glucosendashconfirmed symp-tomatic hypoglycemic episodes during thetrial The EAC-confirmed cause of deathwas infection for one patient cause ofdeath was undetermined for the remain-ing two patients because their medicalrecords were unavailable

Compared with placebo pulse rateincreased for the oral semaglutidearms with ETD of 2ndash4 beatsmin at week26 (all groups P 005) and 1ndash2 beatsmin at week 52 (P 005 for oralsemaglutide 14 mg only) while on treat-ment There were no clinically relevantchanges in laboratory safety parametersor other vital signs reported in any pa-tients (Supplementary Table 9)

CONCLUSIONS

In this trial oral semaglutide 3 7 and14 mg provided dose-dependent statis-tically significant reductions in HbA1c and

body weight compared with placebo over52 weeks in patients with type 2 diabetesinadequately controlledwith insulinwithorwithoutmetformin Furthermore oralsemaglutide treatment enabled up to542 of patients to achieve HbA1c70 (53 mmolmol) at week 52 (treat-ment policy estimand) Better glycemiccontrol was achieved with oral semaglutide7 and 14 mg compared with placebo atweeks 26 and 52 despite lower total dailyinsulin dosages relative to baseline Thesefindings support the addition of GLP-1RAsas an effective treatment intensificationstrategy for patients who are unable toreach or maintain HbA1c targets withinsulin alone (17) as recommended incurrent treatment guidelines (3)

TheHbA1c and bodyweight reductionswith oral semaglutide in this trial weresimilar to those reported in otherPIONEER trials (1318ndash20) Typical of apopulation on established insulin therapypatients in PIONEER 8 were older and had

Table 1mdashKey supportive secondary end points

Treatment policy estimand Trial product estimand

Oral semaglutide Oral semaglutide

3 mg 7 mg 14 mg Placebo 3 mg 7 mg 14 mg Placebo

Patients n 184 182 181 184 184 182 181 184

HbA1c 70Week 26Patients meeting end point n () 50 (284) 74 (425) 101 (584) 12 (68) 50 (309) 70 (446) 96 (658) 11 (68)EOR vs placebo 561 1237 2252 d 635 1421 3184 d

95 CI 277 1137 612 2500 1114 4551 d 310 1300 689 2929 1535 6604 d

P value 00001 00001 00001 d 00001 00001 00001 d

Week 52Patients meeting end point n () 50 (289) 67 (396) 91 (542) 16 (93) 37 (356) 53 (469) 72 (643) 10 (101)EOR vs placebo 402 721 1296 d 459 790 1600 d

95 CI 213 758 384 1354 691 2432 d 222 950 382 1636 777 3291 dP value 00001 00001 00001 d 00001 00001 00001 d

Seven-point SMBG (mmolL)Week 26Estimated mean 89 84 81 97 88 82 77 97Estimated change from baseline 211 217 219 203 212 217 223 203ETD vs placebo 208 214 217 d 209 215 221 d

95 CI ndash13 ndash03 ndash18 ndash09 ndash21 ndash12 d ndash14 ndash05 ndash19 ndash10 ndash25 ndash16 d

P value 00006 00001 00001 d 00001 00001 00001 dWeek 52Estimated mean 85 84 80 92 85 83 79 93Estimated change from baseline 215 216 220 208 215 217 221 207ETD vs placebo 206 208 211 d 208 210 214 d95 CI ndash12 ndash01 ndash13 ndash03 ndash17 ndash06 d ndash13 ndash03 ndash15 ndash05 ndash18 ndash09 d

P value 00161 00035 00001 d 00012 00001 00001 d

Proportions are observed proportions of patients with nonmissing information P values are unadjusted two-sided P values for the test of nodifference Treatment policy estimand ANCOVA for continuous end points and logistic regression for binary end points using data irrespectiveof discontinuation of trial product or initiation of rescuemedication Missing values were imputed by a patternmixture model using multiple imputationPattern was defined by randomized trial product and treatment status Trial product estimand mixedmodel for repeated measurements for continuousend points and logistic regression for binary end points Data collected after discontinuation of trial product or initiation of rescue medication wereexcluded For binary end points missing values were imputed from patients randomized to the same trial product using sequential multiple imputationEOR estimated odds ratio SMBG is reported as plasma-equivalent values of capillary whole-blood glucose


Figure 2mdashBody weightndashrelated efficacy end points A Observed changes from baseline in body weight over time placebo oral semaglutide3 mg oral semaglutide 7 mg oral semaglutide 14 mg B Estimated changes from baseline in body weight C Observed proportions of patientsachieving $5 weight loss D Observed proportions of patients achieving HbA1c 70 without hypoglycemia and without body weight gainTreatment policy estimand ANCOVA for continuous end points and logistic regression for binary end points using data irrespective of discontinuationof trial product or initiation of rescue medication Missing values were imputed by a pattern mixture model using multiple imputation Pattern wasdefined by randomized trial product and treatment status Trial product estimand mixed model for repeated measurements for continuous endpoints and logistic regression for binary end points Data collected after discontinuation of trial product or initiation of rescue medication wereexcluded For binary end points missing values were imputed from patients randomized to the same trial product using sequential multipleimputation Statistically significant ETD or estimated odds ratio (EOR) versus placebo in favor of oral semaglutide P values are unadjusted two-sidedP values for the test of no difference daggerSevere or blood glucosendashconfirmed (31 mmolL [56 mgdL]) symptomatic hypoglycemic episode


more advanced disease than those inthese other trials (1318ndash20) Howeverthe similarities in the results regardless ofpopulation differences highlight the con-sistency of effect of oral semaglutideacross the spectrum of care Furthermorethese clinical benefits are consistentwith results achieved with subcutane-ous semaglutide in patients treatedwith insulin (6) suggesting that theseindividuals could benefit from semaglu-tide regardless of administration routeOral semaglutide may help to over-

come some of the side effects associatedwith insulin use that contribute to thera-peutic inertia in the initiation or intensifi-cationofan insulin regimen(2122) suchasweight gain (23) When added to insulin inthe present trial oral semaglutide resulted

in significant body weight reductions ver-sus placebo Furthermore the total dailyinsulin dosage was significantly reducedfrom baseline with oral semaglutide 7 and14 mg versus placebo at weeks 26 and52 suggesting an insulin-sparing effect atthese doses

Insulin use is also associated with anincreased risk of hypoglycemia (2425)which could be overcome by adding aGLP-1RA Indeed in a prior trial the fixedcombination of liraglutide and insulindegludec improved glycemic controlcompared with the equivalent dose ofinsulin alone without increasing thehypoglycemia risk (26) Similarly inour trial despite the better glycemiccontrol achieved with oral semaglutidethe proportions of patients with at least

one severe or blood glucosendashconfirmedsymptomatic hypoglycemic episodewere similar across treatment arms Forall treatment arms most of these episodesoccurred in patients on basal-bolus insulinThis would be expected from a regimenwith a prandial component and an associ-ation between hypoglycemia and bolus in-sulin has previously been reported (2728)Hypoglycemia is also associated withcardiovascular-related morbidity andmortality (29) In this trial no associa-tion between severe or blood glucosendashconfirmed symptomatic hypoglycemicepisodes and cardiovascular events wasobserved

Consistent with other semaglutide tri-als (6131930) no unexpected safetyissues were identified Gastrointestinal

Table 2mdashOn-treatment AEs and hypoglycemic episodes

Oral semaglutide

3 mg (n 5 184) 7 mg (n 5 181) 14 mg (n 5 181) Placebo (n 5 184)

AEs n () R n () R n () R n () R

Any AE 137 (745) 336 142 (785) 315 151 (834) 344 139 (755) 245

Most frequent AEs affecting $5 of patients in anytreatment arm (by MedDRA preferred term)

Nausea 21 (114) 12 30 (166) 19 42 (232) 37 13 (71) 10Diarrhea 16 (87) 10 22 (122) 15 27 (149) 24 11 (60) 8Decreased appetite 8 (43) 4 18 (99) 11 23 (127) 14 2 (11) 1Vomiting 11 (60) 8 14 (77) 10 18 (99) 18 7 (38) 4Nasopharyngitis 27 (147) 23 21 (116) 18 18 (99) 17 27 (147) 18Upper respiratory tract infection 8 (43) 6 6 (33) 5 13 (72) 8 13 (71) 9Constipation 8 (43) 4 15 (83) 9 12 (66) 8 5 (27) 3Abdominal discomfort 7 (38) 4 11 (61) 7 10 (55) 6 3 (16) 2Urinary tract infection 6 (33) 5 5 (28) 3 10 (55) 6 7 (38) 6Hypertension 3 (16) 2 4 (22) 2 1 (06) 1 11 (60) 6

Serious AEs 25 (136) 23 19 (105) 15 12 (66) 14 17 (92) 13

AEs leading to premature trial product discontinuation 13 (71) 16 16 (88) 17 24 (133) 29 5 (27) 3

AEs leading to premature trial product discontinuationaffecting $3 of patients in any treatment arm (byMedDRA system organ class)

Gastrointestinal disorders 9 (49) 8 12 (66) 13 19 (105) 19 1 (05) 1

Deaths 0 0 0 0 3 (17) 2 0 0

Hypoglycemic episodes by classification andinsulin regimen nN () R nN () R nN () R nN () R

Severe or blood glucosendashconfirmed symptomaticdaggerDagger 52 (283) 105 47 (260) 102 48 (265) 86 54 (293) 82Basal insulin 877 (104) 25 1276 (158) 52 1076 (132) 27 1680 (200) 30Basal-bolus insulin 3671 (507) 206 2973 (397) 168 3170 (443) 155 2772 (375) 155Premixed insulin 836 (222) 92 632 (188) 62 735 (200) 76 1132 (344) 52

SevereDagger 5 (27) 3 1 (06) 1 2 (11) 1 1 (05) 1Basal insulin 0 0 0 0 0 0 0 0Basal-bolus insulin 471 (56) 6 173 (14) 1 170 (14) 1 0 0Premixed insulin 136 (28) 3 0 0 135 (29) 4 132 (31) 3

Data are number and proportion of patients with at least one event or number and proportion of patients experiencing at least one hypoglycemicepisodeover number of patients on each insulin regimenwhere applicable On treatment the period inwhich the patientwas considered treatedwithtrial product MedDRA Medical Dictionary for Regulatory Activities (version 201) R observed rate of episodes per 100 years of exposureHypoglycemic episodes were reported on a separate form from AEs daggerSevere hypoglycemia was defined according to the ADA classification (requiresassistance of another person to actively administer carbohydrate glucagon or other corrective action) DaggerBlood glucose confirmation of symptomatichypoglycemia was based on a blood glucose value 31 mmolL (56 mgdL) with symptoms consistent with hypoglycemia


disorders specifically nausea were themost frequentAEswithoral semaglutidewhich is consistent with other GLP-1RAsand nausea is a known class effect ofthese agents (1) A dose escalation wasused to help to mitigate the occurrenceand severity of nausea and the nauseaevents observed were mild or moderateand of short durationA strength of this trial was the inclusion

of the consecutive insulin dosing stages(capped at baseline levels then fullyadjustable) during the treatment periodThis allowed both the glucose-loweringeffect of oral semaglutide to be deter-mined in a controlled setting and data tobeobtained longer term ina settingmorereflective of clinical practice Howeverthe diversity of insulin types and regi-mens could have limited assessment ofthe interaction of oral semaglutide withspecific regimens In addition titration ofinsulin dosage was performed at thediscretion of individual investigatorsand was not enforced While this was inlinewith the aim of the trial it resulted inHbA1c at week 52 being similar to base-line levels in patients receiving placeboHad the insulin titration after week26 been enforced the comparison be-tween oral semaglutide and placebo withregard to frequency of hypoglycemia andchanges from baseline in insulin dosagecould have been further strengthenedFurthermore while the use of a placebocontrol allowed the evaluation of treat-ment effect using an active comparatorinstead could have provided additionalinsight into the relative risks or benefitsof oral semaglutide comparedwith otheravailable approachesIn summary when added to insulin in

the setting of inadequately controlledtype 2 diabetes oral semaglutide wassuperior to placebo at improving glycemiccontrol and reducing body weight over26 weeks with significant differences alsoseen at 52weeks andwith no increase inthe risk of hypoglycemia Further-more the overall safety profile wasconsistent with that of other GLP-1RAs

Acknowledgments Emisphere is acknowl-edged for providing a license to the EligenTechnology the sodium N-(8-[2-hydroxybenzoyl]amino) caprylate component of oral semaglutideThe authors thank the patients investigators trialsite staff and Novo Nordisk employees involvedin the trial In addition the authors thank SophieWalton of Spirit Medical Communications Group

Ltd for medical writing and editorial assistance(funded by Novo Nordisk AS) and Brian BekkerHansen of Novo Nordisk for reviewing themanuscriptFunding and Duality of Interest PIONEER 8was funded by Novo Nordisk AS DenmarkJBB is supported by grants from the ADA Na-tional Institutes of Health (UL1-TR-002489 U01-DK-098246 UC4-DK-108612 U54-DK-118612) andPatient-Centered Outcomes Research Insti-tute BZ has served on scientific advisory boardsof and received honoraria or consulting fees fromAstraZeneca Boehringer Ingelheim Eli Lilly JanssenMerck Sanofi and Novo Nordisk VRA hasreceived consulting fees from Adocia AstraZenecaBD Novo Nordisk Sanofi and Zafgen andresearch grant support paid to her institutionfrom AstraZenecaBristol-Myers Squibb CalibraEisai Fractyl Janssen Novo Nordisk Sanofi andTheracos and her spouse is an employee ofMerck Research Laboratories JBB has receivedconsulting fees paid to his institution fromAdocia AstraZeneca Dance Biopharm Eli LillyMannKind NovaTarg Novo Nordisk SenseonicsvTv Therapeutics and Zafgen has received re-search grant support from Novo Nordisk Sanofiand vTv Therapeutics is a consultant to CiriusTherapeutics CSL Behring Neurimmune AG andWhole Biome and holds stock options in MellitusHealth PhaseBio Stability Health and WholeBiome BC has received research grant supportfrom Amgen Novo Nordisk Pfizer Sanofi andRegeneron Pharmaceuticals and has served onscientific advisory boards of and received hon-oraria or consulting fees from Abbott AkceaAmgen AstraZeneca Genfit Pierre Fabre EliLilly Merck Sharp amp Dohme Merck amp CoNovo Nordisk Regeneron Pharmaceuticalsand Sanofi SBH has received research grantsupport from AstraZeneca Boehringer Ingel-heim Eli Lilly Janssen Novo Nordisk andSanofi and has served on scientific advisoryboards of provided consultation to and receivedhonoraria or consulting fees from AbbottAstraZeneca Eli Lilly Janssen Merck NovoNordisk and Sanofi EA has served on scientificadvisory boards of and received honoraria orconsulting fees from Alcon Astellas PharmaAstraZeneca Eli Lilly Kowa Pharmaceutical Nip-pon Boehringer Ingelheim NovoNordisk Sanofiand Terumo Corporation has received honorariafor lectures from Astellas Pharma Merck Sharp ampDohme Novo Nordisk Ono Pharmaceutical andSanofi and received scholarship grants fromAstellas Pharma Daiichi Sankyo MitsubishiTanabe Pharma Nippon Boehringer IngelheimNovo Nordisk Ono Pharmaceutical SanofiShionogi Sumitomo Dainippon Pharma andTakeda Pharmaceutical STH KBP andMJT-J are employees of Novo Nordisk thesponsor of this trial No other potential conflictsof interest relevant to this article were reportedAuthorContributionsBZ served as the globalsignatory investigator of the trial BZ VRAJBB BC SBH STH KBP MJT-J andEA were involved in the acquisition analysis orinterpretation of data and assisted with draftingand critical revision of the manuscript STHKBP and MJT-J designed the trial MJT-Jwas responsible for the statistical analysis BZis the guarantor of this work and as such hadfull access to all the data in the study and

takes responsibility for the integrity of the dataand the accuracy of the data analysisPrior Presentation Parts of this study werepresented in abstract form at the 79th ScientificSessions of the American Diabetes AssociationSan Francisco CA 7ndash11 June 2019

References1 Htike ZZ Zaccardi F Papamargaritis D WebbDR Khunti K Davies MJ Efficacy and safety ofglucagon-like peptide-1 receptor agonists intype 2 diabetes a systematic review andmixed-treatment comparison analysis DiabetesObes Metab 201719524ndash5362 Bethel MA Patel RA Merrill P et al EXSCELStudy Group Cardiovascular outcomes withglucagon-like peptide-1 receptor agonists inpatients with type 2 diabetes a meta-analysisLancet Diabetes Endocrinol 20186105ndash1133 American Diabetes Association Pharmaco-logic approaches to glycemic treatment stand-ards of medical care in diabetes - 2019 DiabetesCare 201942(Suppl 1)S90ndashS1024 Arnett DK Blumenthal RS Albert MA et al2019 ACCAHA guideline on the primary pre-vention of cardiovascular disease a report of theAmerican College of CardiologyAmerican HeartAssociation Task Force onClinical PracticeGuide-lines Circulation 2019140e596ndashe6465 CastellanaM Cignarelli A Brescia F Laviola LGiorgino F GLP-1 receptor agonist added toinsulin versus basal-plus or basal-bolus insulintherapy in type 2 diabetes a systematic reviewand meta-analysis Diabetes Metab Res Rev201935e30826 Rodbard HW Lingvay I Reed J et alSemaglutide added to basal insulin in type 2diabetes (SUSTAIN 5) a randomized controlledtrial J Clin Endocrinol Metab 20181032291ndash23017 Pozzilli P Norwood P Jodar E et al Placebo-controlled randomized trial of the addition ofonce-weekly glucagon-like peptide-1 receptoragonist dulaglutide to titrated daily insulin glar-gine in patients with type 2 diabetes (AWARD-9)Diabetes Obes Metab 2017191024ndash10318 Ahmann A Rodbard HW Rosenstock J et alNN2211-3917StudyGroup Efficacy and safety ofliraglutide versus placebo added to basal insulinanalogues (with or without metformin) in pa-tients with type 2 diabetes a randomizedplacebo-controlled trial Diabetes Obes Metab2015171056ndash10649 Porcellati F Lucidi P Bolli GB Fanelli CG GLP-1 RAs as compared to prandial insulin after failureof basal insulin in type 2 diabetes lessons fromthe 4B and Get-Goal DUO 2 trials Diabetes Metab2015416S16ndash16S2010 Buckley ST Baeligkdal TA Vegge A et al Trans-cellular stomach absorption of a derivatizedglucagon-like peptide-1 receptor agonist SciTransl Med 201810eaar704711 Aroda VR Saugstrup T Buse JB DonsmarkM Zacho J Davies MJ Incorporating and inter-preting regulatory guidance on estimands indiabetes clinical trials the PIONEER 1 randomizedclinical trial as an example Diabetes ObesMetab2019212203ndash221012 International Council onHarmonization Sta-tistical principles for clinical trials E9 [Internet]1998 Available from httpswwwichorgproductsguidelinesefficacyarticleefficacy-guidelineshtml Accessed 27 March 2019


13 Rosenstock J Allison D Birkenfeld AL et alPIONEER 3 Investigators Effect of additionaloral semaglutide vs sitagliptin on glycated he-moglobin in adults with type 2 diabetes un-controlled with metformin alone or withsulfonylurea the PIONEER 3 randomized clinicaltrial JAMA 20193211466ndash148014 International Council on HarmonizationEstimands and sensitivity analysis in clinical trialsE9 (R1) [Internet] 2017 Available from httpswwwichorgfileadminPublic_Web_SiteICH_ProductsGuidelinesEfficacyE9E9-R1EWG_Step2_Guideline_2017_0616pdf Accessed 27March 201915 Bretz F Posch M Glimm E Klinglmueller FMaurer W Rohmeyer K Graphical approaches formultiple comparison procedures using weightedBonferroni Simes or parametric tests Biom J201153894ndash91316 Little RJA Rubin DB Statistical Analysis withMissingData NewYorkNY JohnWileyampSons 198717 Davies MJ DrsquoAlessio DA Fradkin J et alManagement of hyperglycemia in type 2 diabe-tes 2018 A consensus report by the AmericanDiabetes Association (ADA) and the EuropeanAssociation for the Study of Diabetes (EASD)Diabetes Care 2018412669ndash270118 Rodbard HW Rosenstock J Canani LH et alPIONEER 2 Investigators Oral semaglutide ver-sus empagliflozin in patients with type 2 diabetes

uncontrolled on metformin the PIONEER2 trial Diabetes Care 2019422272ndash228119 Aroda VR Rosenstock J Terauchi Y et alPIONEER 1 investigators PIONEER 1 random-ized clinical trial of the efficacy and safetyof oral semaglutide monotherapy in com-parison with placebo in patients with type 2diabetes Diabetes Care 2019421724ndash173220 Pratley R Amod A Hoff ST et al PIONEER4 investigators Oral semaglutide versus sub-cutaneous liraglutide and placebo in type 2 di-abetes (PIONEER 4) a randomised double-blindphase 3a trial Lancet 201939439ndash5021 Peyrot M Rubin RR Khunti K Addressingbarriers to initiation of insulin in patients withtype 2 diabetes Prim Care Diabetes 20104(Suppl 1)S11ndashS1822 Allen NA Zagarins SE Feinberg RGWelch GTreating psychological insulin resistance intype 2 diabetes J Clin Transl Endocrinol 201671ndash623 Triplitt C Improving treatment successrates for type 2 diabetes recommendationsfor a changing environment Am J Manag Care201016(Suppl)S195ndashS20024 Erpeldinger S RehmanMBBerkhoutC et alEfficacy and safety of insulin in type 2 diabetesmeta-analysis of randomised controlled trialsBMC Endocr Disord 2016163925 Liu SC Tu YK Chien MN Chien KL Effect ofantidiabetic agents added to metformin on

glycaemic control hypoglycaemia and weightchange in patients with type 2 diabetes a net-work meta-analysis Diabetes Obes Metab 201214810ndash82026 Buse JB Vilsboslashll T Thurman J et alNN9068-3912 (DUAL-II) Trial InvestigatorsContribution of liraglutide in the fixed-ratiocombination of insulin degludec and liraglutide(IDegLira) Diabetes Care 2014372926ndash293327 Holman RR Thorne KI Farmer AJ et al 4-TStudy Group Addition of biphasic prandial orbasal insulin to oral therapy in type 2 diabetesN Engl J Med 20073571716ndash173028 Bajaj HS Venn K Ye C et al Lowest glucosevariability and hypoglycemia are observed withthe combination of a GLP-1 receptor agonist andbasal insulin (VARIATION Study) Diabetes Care201740194ndash20029 Goto A Arah OA GotoM Terauchi Y NodaM Severe hypoglycaemia and cardiovascu-lar disease systematic review and meta-analysis with bias analysis BMJ 2013347f453330 Davies M Pieber TR Hartoft-Nielsen MLHansen OKH Jabbour S Rosenstock J Effectof oral semaglutide compared with placeboand subcutaneous semaglutide on glycemiccontrol in patients with type 2 diabetes a ran-domized clinical trial JAMA 20173181460ndash1470


subcutaneous once-weekly formulationimproved glycemic control and reducedbody weight from baseline when usedalongside basal insulin in patients withtype 2 diabetes (6)An oral formulation of semaglutide

has been developed and is the first oralGLP-1RA to enter phase 3 trials Aspeptides have low oral bioavailabilityoral semaglutide is coformulated withthe absorption enhancer sodiumN-(8-[2-hydroxybenzoyl] amino) caprylatewhichfacilitates semaglutide absorption acrossthe gastric mucosa (10)This article reports the findings of the

Peptide Innovation for Early DiabetesTreatment 8 (PIONEER 8) trial whichinvestigated the efficacy safety and tol-erability of oral semaglutide added ontoinsulin (basal basal-bolus or premixed)with or without metformin in patientswith uncontrolled type 2 diabetes


Trial DesignThis randomized double-blind placebo-controlled parallel-group trial was con-ducted at 111 sites in nine countries(Supplementary Appendix 1) between2 February 2017 and 18 January 2018(NCT03021187) There was a 2-weekscreening period 52-week treatmentperiod and 5-week follow-up period(Supplementary Fig 1A)Patients were randomized 1111 to

once-daily oral semaglutide 3 7 or 14mgor placebo using an interactive web re-sponse system (Supplementary Appendix2) Randomization was stratified by pa-tientsrsquo country of origin (Japanese ornon-Japanese) and background treatment(metformin or no metformin basal basal-bolus or premixed insulin)A 20 reduction in total daily insu-

lin dosage was recommended at ran-domization and maintained to week 8(Supplementary Fig 1B) unless an in-crease was required to prevent acutemetabolic deterioration The treatmentperiod was then split into two stagesdefined by restrictions in total daily in-sulin dosage It could be altered duringweeks 8ndash26 without exceeding the pre-randomization dosage and was freelyadjustable at the investigatorrsquos discretionduring weeks 26ndash52 Throughout thetrial the total daily insulin dosage could bereduced as needed Itwas recommendedthat adjustments were made based onthe lowest of three self-measured blood

glucose (SMBG) values preferably mea-sured on 3 consecutive days prior toeach phone contactsite visit with theaim of obtaining a fasting plasma glucose(FPG) concentration of 40ndash55 mmolL(71ndash99 mgdL) and HbA1c 70 (53mmolmol) (Supplementary Appendix3) In brief dosage was increased inincrements of 2unitsbasedonFPGvaluesstarting at 2 units for 56ndash70 mmolL(100ndash126 mgdL) up to 8 units if 90mmolL (162 mgdL) For patients onbasal-bolus insulin takenmore than oncedaily it was recommended to titrateeach dose separately For patients on basal-bolus insulin recommendations wereprovided only for the basal component

The trial protocol was approved by theinstitutional review boardindependentethics committee at each site and thetrial was conducted in accordance withInternational Council on Harmonization(ICH) Good Clinical Practice guidelinesand the Declaration of Helsinki Allpatients provided written informedconsent

Two different scientific questions re-lated to the efficacy objectives were ad-dressed through the definition of twoestimands the treatment policy estimandand the trial product estimand Bothestimands were defined based on inter-actions with regulatory agencies (11)

The treatment policy estimand ad-dressed the question of the treatmenteffect for all randomized patients regard-less of trial product discontinuation or useof rescue medication This estimand re-flects the intention-to-treat principle asdefined in ICH E9 (12) This estimandreflects the effect of initiating treatmentwith oral semaglutide compared withinitiating treatment with placebo bothpotentially followed by either discontinu-ation of trial product andor addition of orswitch to another glucose-lowering drug

The trial product estimand addressedthe question of the treatment effect forall randomized patients under the as-sumption that all patients remained ontrial product for the entire planned du-ration of the trial and did not use rescuemedication This estimand aims at re-flecting the effect of oral semaglutidecompared with placebo without the con-founding effect of rescue medicationThe statistical analysis that was appliedto estimate this estimand is similar tothat used in the majority of previouslypublished phase 3a diabetes trials (13)

Trial product discontinuation and initi-ation of rescue medication were ac-counted for by the treatment policystrategy for the treatment policy estimandand by the hypothetical strategy for thetrial product estimand as defined in draftICH E9 (R1) (14) Further details are pro-vided in Supplementary Appendix 4

Patient PopulationAdult patients with type 2 diabetesdiagnosed $90 days before screeningwith baseline HbA1c 70ndash95 (53ndash80mmolmol) were enrolled Patients wererequired to be on a stable regimen ofbasal basal-bolus (in any combination)or premixed insulin (including combina-tions of soluble insulin) at$10 unitsdayfor $90 days before screening If usedconcomitant metformin was required tobe at a stable dosage ($1500mg daily orthe maximum tolerated dosage) for$90 days before screening Asidefrom metformin the insulin regimensdescribed above or short-term (14 days)changes in insulin dosage for acuteillness use of any other glucose-loweringmedication was not allowed in the 90days before screening Full eligibilitycriteria are provided in SupplementaryTable 1

Drug AdministrationAs the presence of food or liquid in thestomach impairs absorption of oralsemaglutide (10) patients were instructedto administer trial product in the morningin a fasting state with120mL (4 fl oz)water then towait at least 30min beforethe first meal of the day or taking otheroral medication Tablets were not tobe broken or chewed These instructionswere to ensure sufficient absorption oforal semaglutide Patients randomized tooral semaglutide 3mgwere initiated andremained on the 3-mg dose Those ran-domized to 7 and14mgbegan treatmentat 3 mg and the dose was escalated to7 mg after 4 weeks and to 14 mg after afurther 4 weeks until the randomizeddose was achieved Patients and inves-tigators were blinded to all dose esca-lation steps

Glucose-lowering rescue medication(either newglucose-loweringmedicationor intensification of existing medication)was available to patients taking trialproduct who had persistent or unac-ceptable hyperglycemia based on pre-defined FPG and HbA1c rescue criteria













RESULTS





























CONCLUSIONS








Patients n 184 182 181 184 184 182 181 184


95 CI 277 1137 612 2500 1114 4551 d 310 1300 689 2929 1535 6604 d

P value 00001 00001 00001 d 00001 00001 00001 d


95 CI 213 758 384 1354 691 2432 d 222 950 382 1636 777 3291 dP value 00001 00001 00001 d 00001 00001 00001 d




P value 00161 00035 00001 d 00012 00001 00001 d












Oral semaglutide


AEs n () R n () R n () R n () R

Any AE 137 (745) 336 142 (785) 315 151 (834) 344 139 (755) 245



Serious AEs 25 (136) 23 19 (105) 15 12 (66) 14 17 (92) 13




Deaths 0 0 0 0 3 (17) 2 0 0





























RESULTS





























CONCLUSIONS








Patients n 184 182 181 184 184 182 181 184


95 CI 277 1137 612 2500 1114 4551 d 310 1300 689 2929 1535 6604 d

P value 00001 00001 00001 d 00001 00001 00001 d


95 CI 213 758 384 1354 691 2432 d 222 950 382 1636 777 3291 dP value 00001 00001 00001 d 00001 00001 00001 d




P value 00161 00035 00001 d 00012 00001 00001 d












Oral semaglutide


AEs n () R n () R n () R n () R

Any AE 137 (745) 336 142 (785) 315 151 (834) 344 139 (755) 245



Serious AEs 25 (136) 23 19 (105) 15 12 (66) 14 17 (92) 13




Deaths 0 0 0 0 3 (17) 2 0 0







































CONCLUSIONS








Patients n 184 182 181 184 184 182 181 184


95 CI 277 1137 612 2500 1114 4551 d 310 1300 689 2929 1535 6604 d

P value 00001 00001 00001 d 00001 00001 00001 d


95 CI 213 758 384 1354 691 2432 d 222 950 382 1636 777 3291 dP value 00001 00001 00001 d 00001 00001 00001 d




P value 00161 00035 00001 d 00012 00001 00001 d












Oral semaglutide


AEs n () R n () R n () R n () R

Any AE 137 (745) 336 142 (785) 315 151 (834) 344 139 (755) 245



Serious AEs 25 (136) 23 19 (105) 15 12 (66) 14 17 (92) 13




Deaths 0 0 0 0 3 (17) 2 0 0



























Oral semaglutide


AEs n () R n () R n () R n () R

Any AE 137 (745) 336 142 (785) 315 151 (834) 344 139 (755) 245



Serious AEs 25 (136) 23 19 (105) 15 12 (66) 14 17 (92) 13




Deaths 0 0 0 0 3 (17) 2 0 0


















2262 diabetes care volume 42, december 2019

Documents