Literature search
After duplication removal, 424 articles were screened: 364 were excluded because these were no RCTs but review articles or case reports. Of the remaining 60 records, 43 were considered potentially relevant (Fig. 1). According to the eligibility criteria of our NMA, we considered 15 studies7,19,28–32,20−27. The included RTCs are listed in Table 1.
Table 1
Characteristics of evaluated trials.
Author
|
Year
|
n° pts treated
|
Arms
|
Regrouped Arms
|
Phase
|
PFS
|
Response rates% (n° Responders) Experimental arm
|
Response rates% (n° Responders) Control arm
|
|
|
Experimental arm
|
Control arm
|
|
|
|
HR (95% CI)
|
Median (months)
Experimental arm/control arm
|
|
|
Hillmen
|
2007
|
149
|
148
|
ALEMTUZUMAB vs CHLORAMBUCIL
|
ALEM vs CHL
|
III
|
0,58 (0,43 − 0,77)
|
14,6/11,7
|
83
|
55
|
Hallek
Fischer (update)
|
2010
2016
|
408
|
409
|
FLUDARABINE + CYCLOPHOSPHAMIDE + RITUXIMAB vs FLUDARABINE + CYCLOPHOSPHAMIDE
|
FCR vs FLU + Cy
|
III
|
0,59 (0,50 − 0,69)
|
56,8/32,9
|
90
|
80
|
Goede (A)
|
2014
|
238
|
118
|
CHLORAMBUCIL + OBINUTUZUMAB vs CHLORAMBUCIL
|
CHL-O vs CHL
|
III
|
0,18 (0,13 − 0,24)
|
26,7/11,1
|
77
|
31
|
Goede (B)
|
2014
|
333
|
330
|
CHLORAMBUCIL + OBINUTUZUMAB vs CHLORAMBUCIL + RITUXIMAB
|
CHL-O vs CHL-R
|
III
|
0,39 (0,31 − 0,49)
|
26,7/15,2
|
78
|
64
|
Goede (C)
|
2014
|
233
|
118
|
CHLORAMBUCIL + RITUXIMAB vs CHLORAMBUCIL
|
CHL-R vs CHL
|
III
|
0,44 (0,34 − 0,57)
|
16,3/11,1
|
66
|
31
|
Hillmen
Offner (update)
|
2015
2020
|
221
|
226
|
CHLORAMBUCIL + OFATUMUMAB vs CHLORAMBUCIL
|
CHL-O vs CHL
|
III
|
0,57 (0,45 − 0,72)
|
22,4/13,1
|
82
|
69
|
Burger
Barr (update)
|
2015
2018
|
136
|
133
|
IBRUTINIB vs CHLORAMBUCIL
|
IBR vs CHL
|
III
|
0,12 (0,07 − 0,20)
|
NR/15,0
|
86
|
35
|
Eichhorst
|
2016
|
282
|
279
|
FLUDARABINE + CYCLOPHOSPHAMIDE + RITUXIMAB vs BENDAMUSTINE + RITUXIMAB
|
FCR vs BR
|
III
|
1,63 (1,24 − 2,13)
|
55,2/41,7
|
95
|
96
|
Chanan-Khan
|
2017
|
225
|
225
|
LENALIDOMIDE vs CHLORAMBUCIL
|
LEN vs CHL
|
III
|
0,99 (0,76 − 1,29)
|
30,8/23,0
|
55
|
66
|
Woyach (A)
|
2018
|
178
|
176
|
IBRUTINIB vs BENDAMUSTINE + RITUXIMAB
|
IBR vs BR
|
III
|
0,37 (0,25 − 0,56)
|
NR/43,0
|
93
|
81
|
Woyach (B)
|
2018
|
170
|
176
|
IBRUTINIB + RITUXIMAB vs BENDAMUSTINE + RITUXIMAB
|
IBR-R vs BR
|
III
|
0,40 (0,27 − 0,60)
|
NR/43,0
|
94
|
81
|
Woyach (C)
|
2018
|
170
|
178
|
IBRUTINIB + RITUXIMAB vs IBRUTINIB
|
IBR-R vs IBR
|
III
|
1,06 (0,66 − 1,70)
|
NR/NR
|
94
|
93
|
Michallet
|
2018
|
121
|
120
|
BENDAMUSTINE + RITUXIMAB vs CHLORAMBUCIL + RITUXIMAB
|
BR vs CHL-R
|
IIIb
|
0,52 (0,34 − 0,81)
|
40,0/30,0
|
74
|
75
|
Shanafelt
|
2019
|
354
|
175
|
IBRUTINIB + RITUXIMAB vs FLUDARABINE + CYCLOPHOSPHAMIDE + RITUXIMAB
|
IBR-R vs FCR
|
III
|
0,35 (0,22 − 0,56)
|
|
90
|
77
|
Moreno
|
2019
|
113
|
116
|
IBRUTINIB + OBINUTUZUMAB vs CHLORAMBUCIL + OBINUTUZUMAB
|
IBR-O vs CHL-O
|
III
|
0,23 (0,15 − 0,37)
|
NR/19,0
|
88
|
73
|
Fischer
Al-Sawaf (update)
|
2019
2020
|
216
|
216
|
VENETOCLAX + OBINUTUZUMAB vs CHLORAMBUCIL + OBINUTUZUMAB
|
VEN-O vs CHL-O
|
III
|
0,33 (0,22 − 0,51)
|
41,5/27,0
|
85
|
71
|
Sharman (A)
|
2020
|
179
|
177
|
ACALABRUTINIB + OBINUTUZUMAB vs CHLORAMBUCIL + OBINUTUZUMAB
|
ACA-O vs CHL-O
|
III
|
0,10 (0,06 − 0,17)
|
NR/22,6
|
96
|
82
|
Sharman (B)
|
2020
|
179
|
177
|
ACALABRUTINIB vs CHLORAMBUCIL + OBINUTUZUMAB
|
ACA vs CHL-O
|
III
|
0,20 (0,13 − 0,30)
|
NR/22,6
|
89
|
82
|
Eichhorst
(A)
|
2021
|
231
|
229
|
VENETOCLAX + IBRUTINIB + OBINUTUZUMMAB VS
CHEMOIMMUNOTHERAPY
|
VEN-IBR-O vs
CIT
|
III
|
--
|
--
|
94
|
81
|
Eichhorst
(B)
|
2021
|
229
|
229
|
VENETOCLAX + OBINUTUZUMAB vs
CHEMOIMMUNOTHERAPY
|
VEN -O vs
CIT
|
III
|
--
|
--
|
96
|
81
|
Eichhorst
(C)
|
2021
|
237
|
229
|
VENETOCLAX + RITUXIMAB vs
CHEMOIMMUNOTHERAPY
|
VEN -R vs
CIT
|
III
|
--
|
--
|
93
|
81
|
Munir
|
2021
|
106
|
105
|
VENETOCLAX + IBRUTINIB vs
CHLORAMBUCIL + OBINUTUZUMAB
|
VEN-IBR vs
CHL + O
|
III
|
--
|
--
|
86
|
83
|
A total of 7958 patients were included, while the 16 different treatments adopted are listed below: Acalabrutinib (ACA), Acalabrutinib + Obinutuzumab (ACA-O), Alemtuzumab, Bendamustine + Rituximab (BR), Chlorambucil (CHL), Chlorambucil + Obinutuzumab (CHL-O), Chlorambucil + Ofatumumab (CHL-OFA), Chlorambucil + Rituximab (CHL-R), Fludarabine + Cyclophosphamide + Rituximab (F-C-R), Fludarabine + Cyclophosphamide (FLU-Cy), Ibrutinib (IBR), Ibrutinib + Obinutuzumab (IBR-O), Ibrutinib + Rituximab (IBR-R), Lenalidomide (LEN), Venetoclax + Ibrutinib (VEN-IBR), Venetoclax + Obinutuzumab + Ibrutinib (VEN-IBR-O), Venetoclax + Obinutuzumab (VEN-O).
Both investigators assessed the risk of bias of the included studies by using Cochrane risk of bias tool (Figure S1 and Figure S2)
Pfs Analysis
For PFS analysis we compared 13 trials, with 6583 patients enrolled. We excluded from this analysis two RTCs [Eichhorst et al. (2021) and Munir et al. (2021)], because no data about PFS were found (Table 1). The network plot of direct and indirect comparisons by considering each treatment simultaneously is represented in Fig. 2.
A total of 4 closed loops were generated: “BR/IBR/IBR-R” (RoR 1.035, CI 1.00,1.12), “CHL/CHL-O/IBR/IBR-R” (RoR 1.771, 1.62,1.94), “BR/CHL/CHL-R/IBR” (RoR 2.026, CI 1.84,2.24) and “BR/FCR/IBR-R” (RoR 2.039, CI 1.91,2.18). The first two loops retained low risk of inconsistency (RoR inferior to 2), whereas the last two loops carried a high risk of inconsistency due to RoR major than 2.
According to our analysis, ACA-O was the best treatment option for the front-line therapy of CLL (SUCRA 100%, probable the best 99.9%), while CHL was most likely the worst treatment option (SUCRA 4%, probable the best 0%) (Figure S3).
The difference in PFS between each treatment could be assumed by the network league table, in which pooled effect of each treatment on PFS is represented. Indeed, Figure S4 shows that ACA-O scored a statistically significant benefit in PFS in all diagonal pair-wises.
We also performed two network meta-analyses according to IgVh status to define the best treatment for UN-only CLL patients. When considering the IgVh-unselected PFS network, we excluded from this network 3 trials [Chanan-Khan et al. (2017), Hillmen et al. (2015) and Michallet et al. (2018)] because no HR data were reported. Thus, the analysis included 10 trials, 2184 mutated patients and 2979 unmutated patients.
In UN-only setting, ACA-O scored the best result (SUCRA value 99.2%, probably the best 90%) (Figure S5). It is worth to note that in this subgroup, the difference between ACA-O vs ACA monotherapy was not statistically significant (Figure S6).
In the MU-only subgroup, ACA-O scored the best position (SUCRA value 98,3%, probably the best 86,3) (Figure S7), followed by IBR-O and VEN-O. Notably, the difference in PFS between ACA-O and IBR-O was not statistically significant (HR 0.25, CrI 0.02, 2.66) (Figure S8).
The summary of all SUCRA values for PFS in unselected, mutated and unmutated population are represented in Fig. 3.
Tolerability Analysis
Tolerability network (Figure S9) included 12 trials and enrolled 6436 patients. We excluded 3 trials [Hillmen at al. (2007), Eichhorst et al. (2021), Munir et al. (2021)], because we were unable to calculate the RR due to missing data; indeed, no information about the incidence of AE in the safety population was mentioned in the texts or in their supplementary.
Based on these premises, we found 3 triangular loops and 1 quadratic loops: BR/IBR/IBR-R (RoR 1.094, CI 1.06,1.13), BR/FCR/IBR-R (RoR 1.240, CI 1.21,1.27), CHL/CHO-O/CHL-R (RoR 2.740, CI 2.65,2.84), BR/CHL/CHL-R/IBR (RoR 2.511, CI 2.41,2.62).
According to our analysis, ACA ranked the best position (SUCRA value 100%, probably the best 100%) in the tolerability network score, whereas IBR-O turned to be the most toxic treatment (SUCRA value 0%, probably the best 1%) (Figure S10). Interestingly, IBR monotherapy ranked the 9th position in the network, whereas all combination regimens including Obinutuzumab were weighted by high risk of toxicity (Figure S11).
Orr Analysis
We performed a NMA to rank the different treatments according to ORR. We analysed 14 trials for a total of 6794 patients. We excluded from this analysis one RCT [Eichhorst et al. (2021)], because no stratified information about the specific schedule of CIT standard arm used to compare to VEN based experimental arm was available to date. In the ORR network, four loops were observed: 1) B-R/CHL/CHL-R/IBR (ROR1.209, CI 1.05,1.39), 2) CHL/CHL-R/CHL-O (ROR 1.216, CI 1.12,1.32), 3) BR/IBR/IBR-R (ROR 4.271, CI 3.72,4.91), 4) B-R/FCR/IBR-R (18.900, CI 16.96,21.06). The loops 3) and 4) retained a very high value of inconsistency.
Our analysis showed that VEN-O ranked the first position (SUCRA value 100%, probably the best 100%) (Figure S12, while ACA-O ranked the second position. Of note, the difference between VEN-O and other treatments is statistically significant in every pair-wise comparison (Figure S14).
Mrd Analysis
For the MRD NMA (Figure S14) we included 10 trials [Goede et al. (2014), Hillmen et al (2015), Woyach et al. (2018), Shanafelt et al. (2019), Moreno et al. (2019), Hallek et al. (2010), Sharman et al. (2020), Al-Sawaf et al. (2020), Eichhorst et al. (2021), Munir et al. (2021)], for a total of 4171 patients.
In the MRD network, three closed loops were found: 1) CHL-O/CIT/CI (RoR 1.104, CI 1.00,1.96), 2) CIT/IBR/IBR-R (RoR 15.574, CI 12.27,19.77), 3) CHL-O/CIT/VEN-IBR-O (RoR 290, CI 266,316). Loops 2) and 3) carried a very high grade of inconsistency.
Our analysis showed that VEN-O is the probable best treatment for MRD (SUCRA 90.4%, probable best 36.2%) versus the second best VEN-O-IBR (SUCRA 84.0%, probable best 35.9) (Figure S15); in the direct comparison the difference between VEN-O and VEN-O-IBR is not statistically significant (Odds Ratio 1.44, CrI 0.00,563.57).
Notably, all the regimens containing Obinutuzumab scored the best results in MRD, and the combination VEN-R was not as well efficient in inducing undetectable MRD (Figure S16).
Summary Of Sucras
The summary of all SUCRA values for each reported outcome object of the present study are represented in Fig. 4.