0% found this document useful (0 votes)
16 views17 pagesT2d Clustering - DR Hamed
This study aimed to subtype long-standing type 2 diabetes (T2D) in a non-white Emirati population using unsupervised soft clustering methods. Five distinct clusters were identified, including a new subtype, mild early onset diabetes (MEOD), alongside established subgroups such as severe insulin-resistant diabetes (SIRD) and mild obesity-related diabetes (MOD). The findings suggest that while clustering can provide insights into T2D subtypes, individual patient management may be limited due to overlapping characteristics among clusters.
Uploaded by
Shaheryar HasanCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
16 views17 pagesT2d Clustering - DR Hamed
This study aimed to subtype long-standing type 2 diabetes (T2D) in a non-white Emirati population using unsupervised soft clustering methods. Five distinct clusters were identified, including a new subtype, mild early onset diabetes (MEOD), alongside established subgroups such as severe insulin-resistant diabetes (SIRD) and mild obesity-related diabetes (MOD). The findings suggest that while clustering can provide insights into T2D subtypes, individual patient management may be limited due to overlapping characteristics among clusters.
Uploaded by
Shaheryar HasanCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 17
PLOS ONE
®
Check for
updates,
Giorenaccess
tation Bayou, Fatooq/ Mana
Hassani M, Osama A, Mukbopacyay , ea
(2024 Eiogiesunerjingsubyzes cong
‘tani ype 2 diabetes, PLoS ONE 195):
304096, ory10.137 Vural
one oB04006|
dtr Yoe Gary Arg National eacare Group,
SINGAPORE
Received: November 92028
Aecopt: May 8, 2024
Published: May 28,2028
Copyright© 2024 Sayoumiet a Tiss an open
seoss atl dstrbed under te eso the
Commons Lice, which
aris unrest se, dstrbton, and
‘epoduction inary medium, rode the orginal
‘thor and sauce are crested.
Data Availabilty Statement Arorymzed, otimary
databases willbe aalabl on request But all
ets’ confitental formation wl no be
Funding: This study was supported by an itera
rat MBRU-C4-R62019-6] ava on May
29,2019, ty ha Calle of Meine, Motanines
8 Rashid Univers of Medicine ang Hath
‘Setenes, Dba UAE. Fre suppor was
bane rom Sando A Wala, Grart Number
‘SWARD-22-013 awarded on August 30, 2022
RESEARCH ARTICLE
Etiologies underlying subtypes of long-
standing type 2 diabetes
Riad Bayoumnio'*, Muhammad Farooq, Fatheya Alawadl?, Mohamed Hassanein®,
‘Aya Osama’, Debasmita Mukhopadhyay’, Fatima Abdul’, Fatima Sulaiman’,
Statny Dsouza5', Fahad Mulla', Fayha Ahmed", Mouza AlSharhan‘, Amar Khamis°
1 College of Medicine, Mohammed Bin Rashid Unversity of Medicine and Heath Sciences, Dubal, UAE,
2 Dua Diabetes Centr, Dubai Heath, Dubai, UAE, 3 Endocrinology Department, Dubai Hospital, Dubai
Heath, Dubai, UAE, 4 Pathology Department, Dubal Hospital, Dubai Health, Dubal, UAE
* rad bavouni@mbey ac a8
Abstract
Background
‘Attempts to subtype, type 2 diabetes (T2D) have mostly focused on newly diagnosed Euro-
pean pationts. In this study, our aim was to subtype T2D in a non-white Emirati ethnic popu
lation with long-standing disease, using unsupervised soft clustering, based on etiological
determinants,
Methods
Tho Auto Cluster model in the IBM SPSS Modeler was used to cluster data from 348 Emirati
patients with long-standing T2D. Five predictor variables (fasting blood glucose (FBG), fast-
ing serum insulin (FSI), body mass index (BMI), hemoglobin Ac (HDAC) and age at diag-
nosis) were used to determine the appropriate number of clusters and their clinical
characteristics. Multinomial logistic regression was used to validate clustering results.
Results
Five clusters were identified; the first four matched Ahlavist et al subgroups: severe insulin-
resistant diabetes (SIRD), severe insulin-deficient diabetes (SID), mild age-related diabe-
tes (MARD), mild obesity-related diabetes (MOD), anda fith new subtype of mild early
onset diabetes (MEOD). The Modeler algorithm allows for soft assignments, in which a data
point can be assigned to muttiple clusters with different probabilities. There were 151
patients (43%) with membership in cluster peaks with no overlap. The remaining 197
patients (57%) showed extensive overlap between clusters at the base of distributions.
Conclusions
Despite the complex picture of long-standing T2D with comorbidities and complications, our
study demonstrates the feasiblity of identitying subtypes and their underlying causes. While
Clustering provides valuable insights into the architecture of T2D subtypes, its application to
Individual patient management would remain limited due to overlapping characteristics.
PLOS ONE | itosi/do ora/10.187 joural pone. 0301035 May 26,2028 Var PLOS ONE
‘Subtypes of lng standing type 2 diabetes
The tudes ad no rl study deg, data
calectn ard anys, decision to pubis or
reparation ote manuscript
‘Competing interests: authors dece ta they
‘donot have any facial support or laionsips
‘hat mgt have posed cai of tres ints
sun,
Therefore, integrating simpitied, personalized metabolic profiles with clustering holds
I indicated thatthe predictor variable was associated
‘with an increased risk of the outcome falling into a particular cluster relative tothe reference
category (Cluster).
23.3 HOMA assessment. FG, FSI and other chemistry assays were performed using a
‘Cobas 6000 Analyzer (Hoffmann La Roche Diagnostics, CA, US). Insulin resistance (IR) and
{cell dysfunction (B) were evaluated by homeostatic model assessment for IR (HOMA-IR)
and prcel dysfunction (HOMA-B) [30]. The HOMA indices were derived from FBG and fast-
{ng serum insulin levels using the following equations:
insulin
HOMA — IR = FBG x a)
01 oe @)
HOMA —~ B = 20 x lin @)
BG °
‘The higher the HOMATR, the greater the peripheral resistance to insulin, while the lower
the HOMA-B, the greater the B-cell dysfunction, Generally, a HOMA-IR value < 1 indicates
‘optimal insulin sensitivity. Levels above 1.9 indicate early resistance; levels above 2.9 indicate
significant resistance, A HOMA-B value < 100 indicates B-cell dysfunction,
3 Results
3.1 Demographics
‘The mean age of the 348 Emirati patients with T2D was 56 years, and the mean duration of
diabetes was 14 years. The mean BMI was 31 and the mean age at diagnosis was 42 years, Gen
der-wise demographic characteristics are shown in Table 1
‘Owing to the long duration and chronicity of T2D, considerable deterioration in the meta-
bolic profiles of selected patients was observed. Of all patients, 90 (26%) had HOMA-IR > 3.0,
indicating peripheral insulin resistance, while 140 (40%) had HOMA-B < 100, indicating.
Table 1. Demographic characeretis of 38 EmiratiT2D patients selected for subtyping ofthe disease
Men (S = 167)
Age (ears) 5634 981)
BMI (Kg) 3019 6.19) 532.26 (600)
Waist hip ato 10006) 094,008)
‘Age at agnosis (years) 41.43 (1032), 42.34 (11.00)
Duration of diabetes (ear) 1491 (10) 13.96.18)
Allvaluesare shown
Mean (Standard Deviation).
ips or/10.197Voual pone 309096 00
PLOS ONE | itosi/do ora/10.187 joural pone. 0301035 May 26,2028 SIT PLOS ONE ‘Subtypes of lng standing type 2 diabetes
“Table2, Prevalence of comorbidities and complications ofT2D in 348 Emir patient recruited for subtyping ofthe disease,
Fron os) ‘Men = 167) women (= 1) Pre
tino i780 soon 008
cats 300) 2000 se
Glascoma Mo +0) 60) oa
cx 803 a0 203 esi?
caD 5109) 350 «6 001
itor of Sake 2609 set) 018 ut
Hyperion 216) m6) 0160) ene
PAD ua +9) 60) on
erigheral Newopahy 15 5) 83) 10166 ain
(CKD- chronic kidney disease; CAD- coronary artery disease; PAD- peripheral artery disease
Allvalues are shown ae Number of patents (percentage)
ied or/0 197 ural pone 1904036 002
pancreatic secretion dysfunction. The remaining 118 (34%) exhibited both pathophysialogical
dysfunctions. The prevalence of comorbidities and complications observed were also high,
with hypertension at 62%, peripheral neuropathy at 53%, retinopathy at 33%, and coronary
artery disease at 15% (Table 2). In most patients, a least two comorbidities or complications of
diabetes were observed,
3.2 Cluster analysis
Results ofthe cluster analysis ofthe cohort of 348 Emirati patients with T2D with long-stand-
ing disease, are shown in Table 3. No significant differences in cluster results were observed
between male and female T2D patients. Therefore, results were reported for the total cohort
throughout the manuscript. Five Clusters were identified in this study. The first four matched
Ablgvist et al (11) subgroups. Cluster 1 had severe insulin-resistant diabetes (SIRD) in 8% of
patients. Cluster 2 had severe insulin deficient diabetes (SIDD) in 16%, Cluster 3 had mild age~
related diabetes (MARD) in 25%. Cluster 4 had mild obesity-related diabetes (MOD) in 21%.
A fifth new subtype of mild early onset diabetes (MEOD) was identified in 30% of most lean
patients,
However, there was extensive overlap between clusters, Cluster 1 (SIRD), with a positive
average silhouette score, did not significantly overlap with any of the other clusters. The other
‘Table3, Resls of clustering analysis of 88 Emirati T2D pati
sing IBM SPSS modeler,
Chusters Number of patients N (0) ‘Average Silhouette Score
‘Toul ‘with no overlap between clusters [with overlap between clusters
siRD Ze) 250) 20) one
sipp s7a6) soa) 176) oz
MARD 8605) 240) us) 0275
‘Mop ma) 23a sous) ais
MEOD, 105 (20) sean) sus) -oni0
Total 348 (100) 15143) 197 (7)
SIRD- severe ins
cay-onset diabetes.
ASilnouette Index of 1.0 indicates no ovelap and <1.0 indicates overlap between clusters.
resistant abet, IDD- severe insulin deficient diabetes; MARD- mil age-related diabetes; MOD- mild obesity-related diabetes: MEOD- mild
Allvaler are shown as Number of patents (percentage)
ne or/¥0.197 oul pone 004036 008
PLOS ONE |ips:/dol org 10.187 Mowrnal pone O204098 May 28, 2024 eri7 PLOS ONE ‘Subtypes of lng standing type 2 diabetes
a@Cluster1 mCluster2 mClusterS mCluster4 BClusterS BOverlap
Fig 1. Distribution of five clusters among 348 Emirati T2D patients, with and without overlap: Distribution of
patents is shown as percentage of total patients (N ~ 348) into overlapping clusters in grey, and percentage of|
patents exclusive to I cluster as ight be (Custer I SIRD), red (Cluster 2-SIDD), dark blue (Cluster 3~
-MARD), green (Cluster MOD), and yellow (Cluster 5- MEOD).
tps or/0.157 our poe 0304036 01
four clusters, with negative average silhouette scores, seemed to overlap extensively. There
‘were 151 patients (43%) with membership in cluster peaks with no overlap, as confirmed by a
Sihouette Index and Bayesian probability of 1.0 (Lable3). The remaining 197 patients (57%)
showed extensive overlap between clusters confirmed by @ Silhouette Index and Bayesian
probability of <1.0(S1 Table) with individuals appearing in two or more clusters (Fig 1).
Principal component analysis (PCA) was used to visualize the dataset and identify the five
“T2D clusters, sit transformed the dataset into alower- dimensional space where the clusters,
were more easly separated (Fig 2). Multinomsial logistic regression was used to explain the
t
fir
f
ig. Display of principal component analysis of clusters of 348 EmiraiT2D patients: Each cre represents a
patlent ina cluster Th colors epson ach cluster as ght blue (Chistes 1-SIRD), red (Chistes 2 SID), dark blue
(Chster 3 MARD),gren (Cluster 4- MOD), and yellow (Cluster S- MEOD),
ie or 17 oual pone conic
PLOS ONE | ins:/doiorg/10.137 Vow pone 0304095 May 28,2024 77 PLOS ONE ‘Subtypes of lng standing type 2 diabetes
‘Tabled, Estimation ofthe contrtation ofthe five independent variables (Age at diagnosis, body mass index, ating insulin, fting blood glucose and HDA) at
predictors for T2D clusters using mutino ral log regression,
‘loser [Pedi RereonCatcint () aoe dae Rats RCH
sino Posing an 0904 0001 247 [146-36]
Mt ovat 0001 2 (tists
Agta 038s 0001 vasa 172]
mG oon 054 1a ase]
bate ozo ozs aaa 34]
spp hate as 2001 Tnaa 12-4529)
Mm aazs ote 13a(.-1 73)
tsa oan 000i 135 (u-L8)
6 oon Soon ‘iofi0e-114
Agee dag se ozs ast aze-106)
MAND ‘Aare qos ons 0001 Lest 18)
aM czas 00s 124.013]
tna ate 0001 fut
HG aout ast 99 (007-01)
baie are 5 045 020-10)
Mop su ease 2001 iene 285)
tsi ous oa, jut
bate cass oss? 130067238
geo ng own ses L2(09-110
Ea ons 0366 ton a9s-105
SIRD. severe inslin-resstant diabetes, SIDD- severe insulin deficient diabetes; MARD- mild age-related diabetes MOD- ld obesity-related diabetes, BMI- body
‘mass index: FBG- fasting blood glucose; HDALc- Hemoglobin Ale
The predictors wth highest contribution ae talcied for each caster.
is or 17 ual poe 908036
relationship between predictor variables and categorical outcomes (clusters) and validate clus-
tering results, We identified predictor variables that were significantly associated with the clus-
ters and quantified the strength of these associations, The higher the regression coefficients (B)
and the odds ratio (OR), the stronger the contribution to the luster (lable 4) The model coef-
ficients ranked the importance ofthe predictor variables in each cluster [S2 Table]
3.3 Cluster characteristics
‘The pathophysiological characteristics, and laboratory data of the five clusters identified in the
cohort of 348 Emirati patients with long-standing T2D is shown in Table 5. Characteristics of
4 clusters matched that of Allqvist et al [11] subgroups. Cluster 5 patients had a novel subtype
‘of mild early-onset diabetes (MEOD) in mostly lean patients.
3.4 Etiological processes governing T2D subtypes
‘To highlight the major etiological processes governing membership of subtypes, we selected
data of patents at the non-overlapping apices of cluster distributions (N = 151), confirmed by
a Silhouette Index > 1.0 (Table).
In Cluster 1 (SIRD), the primary dysfunction was a markedly increased insulin resistance
associated with moderate obesity. The patients had a normal insulin secretory capacity and
moderately abnormal glucose homeostasis. Most of the patients had peripheral neuropathy
Table)
PLOS ONE |ips:/dol org 10.187 Mowrnal pone O204098 May 28, 2024 Bni7 PLOS ONE ‘Subtypes of lng standing type 2 diabetes
‘TableS."Thepathophyriologial characterstis of 2D subtype in Emirai patients with longstanding dscae in Dabs, UAE.
|SIRD N= 27 68) SIDDN 57 (60) | MARDI - 8658) MOD N-=7 (218) | MEODN = 105 (OR) | ToaIN= 981008)
Meera S203) 324097) 006) S46) 510000) 563.008)
Agestdagoris Gan) sla) 79080 1080) en) 365002) #19007
Duration diabetes) 306 vas.@2) (63 160 rsa Mae
‘BMI (Kg/m”) 343 (4.1), 32.0(64) 2993.7) 37.0 (48) 27.232) 3135.7)
Homa wos) 7549) 505) 300) 2708 514)
Homa 226) a9 way 108 108) sau)
8G mid) 15209) 216) 300 ba 509 vs)
ARG (ng) 15 02 za) 1408) 1400 16809) 17340
nin i ‘aio 131) 139669) Bac) 5060 vw00068)
icon 7a) i943) 6208) zat) 7109) 26012)
ood Ure ng) 26) 280) 3509, a0 200 3108,
Serum retininemgdl) 0800020) —a79(048) | ava 078030) ors 020 0791000
coFR min) 9548) 10105) 6c) 603) aun 954
{rine Albumin og) 068 10039 mar a 2166 73.002
Une ceatnine mal) 1367 M70 1856 1080) 08, ven
Alumin‘retinin Ra 520380 tom) 18068 1916) 35000) 76%)
ond Chleerl ng) 151 G6) 169 1809) es) 156099 ro)
DL oil) 50 809 a0 cn ean 703)
LL mg) men 08 00 05) tar 800
‘Triglycerides (mg/dl) 158 (89) 155 (82) 128 (61), 143 (89) 125 (79) 137 (79)
Aram 200 00 260 a0 2G 2a)
asraun) 2) 300, 20) 20) 209 2307
aurauiny nen nan mao 200 neo m8
ln deficient dshetes; MARD- mild age elated diabetes MOD: mild obesity related diabetes MEOD: mid
crly-ontet diabetes BMI- body mat index; HOMA.IR- homeosLatc model siesment for inulin resistance; HOMA.B: homeosaie model asetament of cll
nction: FBG: fasting blood glucose; RBG. random blood glacose: HbAle: Hemoglobin Alc; cGFR estimated glomerla Slraon ate: HDL high density
poprten: LDL- low-density lpopeoti ALT- alanine aminotransferase; AST- aspartate aminotanserase; ALP- Able Phosphatase
‘Alves ae shown as Mean (Standard Deviation,
nis on/¥0 197 ual pon 204095 005
“Table6, Data ofthe major
ea proceses governing T2D subtypes in sobtets of individuals at the non
