The increased or inappropriate expression of genes with oncogenic properties through specific chromosome translocations is an important event in the pathogenesis of B-cell lymphoproliferative diseases. Recent studies have found deletions or translocations of chromosome 7q to be the most common cytogenetic abnormality observed in SLVL, a leukemic variant of SMZL, with the q21 – q22 region being most frequently affected. In three patients with translocations between chromosomes 2 and 7, the cloning of the breakpoints at 7q21 revealed that each was located within a small region of DNA 3.6 kb upstream of the transcription start site of cyclin dependent kinase 6 (CDK6). In each case the translocation event was consistent with aberrant VJ recombination between the immunoglobulin light chain region (Ig kappa) on chromosome 2p12 and DNA sequences at 7q21, resembling the heptamer recombination site. The t(7;21) breakpoint in an additional patient with splenic marginal zone lymphoma (SMZL), resided 66 kb telomeric to the t(2;7) breakpoints juxtaposing CDK6 to an uncharacterized transcript. In two of the SLVL patient samples, the CDK6 protein was found to be markedly over expressed. These results suggest that dysregulation of CDK6 gene expression contributes to the pathogenesis of SLVL and SMZL.
Structural abnormalities of chromosome 7q are common in myeloid malignancies but rare in lymphoid tumors (Fischer et al., 1997; Dascalescu et al., 1999). Recently, several studies have noted an association between the 7q abnormalities and a subgroup of chronic B cell lymphoproliferative disorders in which the tumor cells show lymphoplasmacytoid features (Oscier et al., 1993, 1996; Wong et al., 1998). The histological or hematological diagnoses in these cases include lymphoplasmacytic lymphoma, splenic marginal zone lymphoma (SMZL) and its leukemic variant, splenic lymphoma with villous lymphocytes (SLVL) (Jaffe, 1997). We have previously found 7q abnormalities to be the most common cytogenetic abnormality in SLVL, occurring in 26% of cases (Oscier et al., 1996). Most of the abnormalities comprise deletions involving bands 7q22 and/or 7q32 (Offit et al., 1995; Hernandez et al., 1997a; Dascalescu et al., 1999), or translocations between 7q21 or q22 and other partner chromosomes (Oscier et al., 1996). Occasional cases show more distal deletions of 7q34 – q36 (Hernandez et al., 1997b).
A number of genes of potential importance in malignancy have already been mapped to chromosome 7q. These include the possible tumor suppressor gene CUTL1 (Zeng et al., 1997), the asparagine synthetase gene (ASNS) (Heng et al., 1994) the cyclin dependent kinase 6 gene (CDK6) Bullrich et al., 1995), located at 7q21/22, the fragile site FRA7H at 7q32 (Mishmar et al., 1998), the apoptosis regulatory gene NEDD2 at 7q34 – q35 (Kumar et al., 1995) and the DNA repair gene XRCC2 at 7q36.1 (Tambini et al., 1997). However the genetic consequences of 7q abnormalities in both myeloid and lymphoid tissues remains unknown.
The purpose of this study was to characterize the translocation breakpoints in four cases of SLVL or SMZL with translocations of 7q21 – 22, in order to define the genes involved.
Three patients with SLVL and one patient with SMZL have been investigated, see Table 1 for clinical details. Fluorescent in situ hybridization (FISH) using probes specific for 7q identified a yeast artificial chromosome (YAC) clone, HSC7E947, which spanned the translocation breakpoint in each case. All other YACs analysed showed the appropriate signals either proximal or distal to the translocation breakpoints. To determine the precise site of the breakpoints a high resolution physical map was constructed (Figure 1). It was possible to demonstrate, by FISH, that bacterial artificial chromosome (BAC) H_NH0029A11 also spanned all four breakpoints. Cosmid cos130a6 encompassed the SLVL breakpoints (Figure 2) while cosmid BMTH9, approximately 50 kb distal to cos130a6, spanned the P4 breakpoint.
DNA sequence analysis of cos130a6 showed that it contained the first exon and promoter region of the gene for CDK6. A large genomic region flanking cos130a6 had previously been sequenced allowing us to determine that CDK6 contained 7 exons spanning 218.449 nucleotides of DNA in the 7cen-3′-CDK6-5′-7qter orientation at 7q21.2 (Figure 1). Analysis of this sequence data indicated that CDK6 was the only apparent transcriptional unit in the vicinity of the breakpoints (Figure 1). The CALCR gene was the closest characterized gene telomeric to CDK6, being over 800 kb away (an annotated version of the complete 7q21 region is posted at http://www.genet.sickkids.on.ca/chromosome7/). More information on the genomic clones is also available at this WWW-site. The position of the translocation breakpoints in each patient, as determined by FISH and DNA sequence analysis of the cloned breakpoint, is shown.
The translocation breakpoints in P1, P2, and P3 were mapped precisely within a 1.65 kb region, 3.6 kb upstream of the transcriptional start site of CDK6, and in P4 the breakpoint was located 66 kb upstream of CDK6 (Figure 3). Genomic probes (p1303.0 and p1305.0 from cos130a6 and p67tel from cosmid BMTH9) were used to identify the genomic rearrangements. Probes p1303.0 and p1305.0 identified rearranged bands with enzymes XbaI, SacI and EcoRI in P1 (Figure 3a – c respectively) while P2 showed similar rearrangements (Figure 3b,c). Probe p1305.0 detected different rearrangements in P3 when hybridized to EcoRI and PvuII digested patient DNA (Figure 3d,e respectively). Rearrangements in P4 were not found with probes p1303.0 or p1305.0 but were identified with the more telomeric probe p67tel on XbaI (Figure 3f) and BamHI (not shown) digested patient DNA. No genomic rearrangements of CDK6 were found in an additional 15 SLVL patients who did not have karyotypic abnormalities of 7q21. Breakpoints from the four patients were subsequently cloned and sequenced (Figure 4). In the three patients with SLVL the breakpoint on 2p12 was localized to the heptamer recombination recognition sequence of the kappa light chain variable region genes (L24a in P1 and P2 and the B1 pseudogene in P3 (Lorenz et al., 1988). The 7q21 breakpoint in P1 and P2 was 2064 bp upstream of the transcription start site of CDK6 while the breakpoint in P3 was 1.65 kb telomeric to the P1/P2 breakpoint. In all three cases the 7q21 breakpoint showed DNA sequence identity with the immunoglobulin recombination recognition sequence (five out of seven bases in P1 and P2 and six out of seven bases in P3), (Figure 4). This suggests that all three translocations may have resulted from aberrant VJ recombination. The 7q21 breakpoint in P4 occurred 300 bp into a 25 kb region composed primarily of L1 repetitive elements (Figure 3). The sequence at the breakpoint again showed six out of seven bases of sequence identity with the immunoglobulin recognition sequence. Additionally a transcript on chromosome 21, localized by PCR using a chromosome 21 specific somatic cell hybrid sample, has been identified, within 1 kb of the breakpoint. Sequence analysis identified the presence of a THE-7 element located between the P3 and P4 breakpoints 29 kb telomeric to CDK6 (Figure 1). This site has been previously reported by Wahbi et al. (1997) to be juxtaposed to the Ig heavy chain locus on chromosome 14q32 in a patient with a chronic B-cell disorder and a t(7;14)(q22;q32), (shown in Figure 1 as t(7;14)), however, the proximity to the CDK6 locus was not noted.
To demonstrate the involvement of CDK6 in SMZL we performed RT – PCR, Western blotting and flow cytometry on the tumor cells of P1 and P2. Material was not available for expression studies from P3 and P4. In both P1 and P2 there was over expression of CDK6 compared to SLVL tumor cells from patients lacking the 7q21 rearrangement (Figure 5). The protein had the expected electrophoretic mobility suggesting it was normally translated. The quantity of expressed protein was approximately an order of magnitude lower than that seen in Jurkat cells, which have previously been described as demonstrating high levels of CDK6 expression (Szepesi et al., 1994), this still represents a gross over expression of CDK6 compared to other SLVL cells lacking chromosome 7q21 rearrangements or normal quiescent B or T lymphocytes. Tumor lymphocytes from an additional eight SLVL and 11 chronic lymphocytic leukemia (B-CLL) patients were analysed by flow cytometry and showed no increased level of CDK6 expression.
Dysregulation of oncogenes by translocation to an immunoglobulin heavy or light chain locus is a frequent event in the pathogenesis of B-cell malignancies (Korsmeyer, 1992; Rabbitts 1994; Chesi et al., 1997). Translocations may result from aberrant VDJ or class switch recombination, or arise as a consequence of somatic hypermutation (Goossens et al., 1998) or recombinase activating gene (RAG) mediated transposition (Hiom et al., 1998). This current study implicates the over expression of CDK6 in the pathogenesis of a B cell malignancy. In the three SLVL cases, the CDK6 gene is juxtaposed to the V region of the kappa light chain locus. Immunophenotypic and VH gene analysis is consistent with the target cell arising from a memory B cell which has traversed the germinal centre and undergone somatic hypermutation but not immunoglobulin class switching. The translocations involving the kappa gene could be a result of aberrant VDJ recombination during B cell development, but could also be a consequence of immune receptor editing in germinal centre cells expressing the recombinase activating genes RAG1 and RAG2. A different mechanism must be proposed for the fourth case (P4) in which the translocation involves an uncharacterized expressed transcript on chromosome 21. In this case, like that of the t(7;14), mapped in this report, the translocation involved elements close to the end of a repetitive genomic element; in (P4) this was an L1 repeat while a THE-7 element was the site of the t(7;14) translocation.
CDK6 together with CDK2 and CDK4 are a group of serine/threonine kinases which positively regulate the G1 to S phase transition by phosphorylating the retinoblastoma protein (pRb), an event which leads to the release of E2F transcription factors that control S phase genes (Easton et al., 1998). The activity of the G1 cyclin dependent kinases is positively regulated by binding to a D type cyclin and the CDK activating enzyme CAK, and negatively regulated by two families of CDK inhibitors which include the INK4 group (p15, p16, p18 and p19) and the Cip/Kip group including p21 and p27.
The majority of human tumors have genetic abnormalities that dysregulate components of the RB1 cell cycle pathway. A recent study has investigated CDK6 expression in both normal and neoplastic lymphoid tissues using immunocytochemistry and flow cytometry (Chilosi et al., 1998). CDK6 was strongly expressed in both normal cortical thymocytes and T cell acute lymphoblastic leukemia or lymphoblastic lymphoma. CDK6 expression in these tumors was not due to gene amplification, or associated with genomic rearrangements. CDK6 was either not expressed or expressed at a very low level in a variety of other lymphomas including peripheral T cell non Hodgkin's lymphoma (NHL) and both high grade and low grade B NHL, including two cases of mantle cell lymphoma and one case of lymphoplasmacytoid lymphoma. In addition to T cell lymphomas, increased activity of CDK6 has been demonstrated in a variety of other malignancies including squamous cell carcinoma (Timmermann et al., 1997), gliomas (Costello et al., 1997) and in a neuroblastoma cell line (Easton et al., 1998). The mechanisms responsible for the overactivity include genomic amplification of 7q21 including CDK6 (Costello et al., 1997), and a CDK6 mutation which disrupts the binding of p16 to the CDK, thereby preventing inhibition of CDK6 protein kinase activity (Easton et al., 1998).
The data presented here demonstrate a novel mechanism for increased CDK6 expression in four cases of SLVL/SMZL. The two cases in which a t(2;7) translocation was the sole chromosomal abnormality have pursued a benign clinical course whereas the cases in which a 7q translocation was part of a complex karyotype had progressive disease (Table 1). Particularly in patients P1 and P2, where no other karyotypic abnormality was found, it seems likely that CDK6 overexpression is implicated in the pathogenesis of SLVL/SMZL. However, the exact role played by such increased expression remains speculative.
Flow cytometric analysis of the proliferation marker Ki-67, with simultaneous measurement of DNA, in six SLVL cases in this study, showed a small number of circulating tumor cells, typically <2% in G1, <1% in S phase, and the great majority in G0 (results not shown), indicating that SLVL is a disease with a low proliferative index. The common clinical pattern of slow accumulation of circulating lymphocytes is consistent with subtle alterations in the balance between cell proliferation, differentiation, and apoptosis. Increasingly, cell cycle regulators are described as having a role in more than one of these functions (Latham et al., 1996; Kaldis et al., 1998; Schneider et al., 1998).
We investigated 15 additional cases of SLVL without abnormalities of chromosome 7q and failed to demonstrate genomic rearrangements of CDK6; the eight cases investigated for increased protein expression were also negative. Investigation of other regulators of the G1 to S transition in SLVL is clearly indicated. Additionally, a wider study of CDK dysregulation in other low grade B cell malignancies may be productive, particularly since drugs which inhibit cyclin dependent kinases are currently under development (Konig et al., 1997). This study demonstrates a novel mechanism for increased CDK6 expression, which may function in an oncogenic manner, and provides direct evidence that dysregulation of key elements of the G1 stage of the cell cycle is implicated in the pathogenesis of SLVL/SMZL.
Materials and methods
Primary tumor samples
Samples were obtained by informed consent from patients of the Royal Bournemouth General Hospital and the Truro and Trelisk General Hospital, UK. Tumor cells were isolated by standard methods (Corcoran et al., 1998). Data on patients P1 – P4 have been presented in other studies (Oscier et al., 1996).
Cytogenetics and FISH
Karyotype analysis was carried out in the Bournemouth Cytogenetics Laboratory on TPA (tetradecanoyl phorbol 12 myristate 13 acetate) stimulated blood cultures, using a GTL (Geimsa, trypsin, Leishman) banding technique. Fresh slides were prepared for FISH analysis from stored fixed cell suspensions. YAC HSC7E947, BAC H_NH0029A11 and cosmid cos130a6 were from a chromosome 7-specific YAC (Kunz et al., 1994), a total genomic library, and the Lawrence Livermore National Laboratory chromosome 7 specific cosmid library, respectively. Cosmid BMTH9 was from a genomic library constructed from YAC HSC7E947 in this study. Probes were labeled with digoxygenin (Roche Diagnostics) by nick translation and detected by a sheep antibody conjugated with rhodamine (Roche Diagnostics). They were used in conjunction with biotinylated whole chromosome 2 or 7 paint (Cambio, Cambridge, UK) which was detected with fluorescein avidin (Vector Laboratories, UK), in dual color FISH experiments. Metaphases were examined using a Zeiss Axioskop epifluorescence microscope with a 100 W mercury vapor lamp. Images were captured with a cooled CCD camera (Photometrics, Tucson, AZ, USA) run by Smartcapture software (Vysis, UK).
Southern blot analysis
The probes used to detect the breakpoints included p1303.0; a 3 kb EcoRI fragment isolated from cosmid cos130a6 which contains the first exon of CDK6, p1305.0; a 5 kb EcoRI fragment also isolated from cos130a6 which is located upstream of the start of transcription of CDK6. To identify the t(7;21) breakpoint probe p67tel (668 bp) was generated by PCR using primers P67for; 5′-CCATGTTGCTGTCCCTC-3′, and P67rev 5′-TTTGGGTAAGCCAAGAGATAATGG-3′, and is located 67 kb telomeric to the start of transcription of CDK6.
Translocation breakpoint fragments
The cloned chromosome 7 breakpoints included re-arranged SacI fragments from patients P1 and P2 using nested primers CD13; 5′-CCGCAGCGAAGGAAGCGTCG-3′, and CD14; 5′-TTCGGCGGAGAGAAACGGGAGCAGC-3′, a rearranged SacI fragment cloned from patient P3 using nested primers CD17; 5′-ATTAGTGGGAGAGAGTGGGACG-3′ and CD18; 5′-TCGGCGGAGTTTCCAAACCCAG-3′, and a rearranged XbaI fragment cloned from patient 4 using the nested primers P67rev; 5′-TTTGGGTAAGCCAAGAGATAATGG-3′ and P67rev2; 5′-CTAGGAGCACATGCTTGACC-3′. All tumor DNA samples were digested with the appropriate restriction enzyme before ligation to prepared pBluescript SK (Stratagene), predigested with the same enzyme and alkaline phosphatase treated. Nested PCR was then performed with the afore-mentioned primer sets, using the M13 reverse primer and T3 primers as linker primers and the resulting PCR products cloned into pGEM vector (Promega UK).
RT – PCR assays
2×105 CD19-RPE-Cy5 (Dako) positive B lymphocytes were collected by FACS sorting using a Beckton Dickinson FACSCaliber flow cytometer. RNA was recovered and cDNA synthesized by standard methods (Chesi et al., 1997). The efficiency of cDNA synthesis was monitored by measuring incorporation of labeled dCTP, and the quality of the cDNA verified by PCR amplification using RNA specific primers for a 520 bp fragment of the β2-microglobulin gene. CDK6 cDNA was amplified under standard conditions for 23 cycles using primers CDK6RTA; 5′-GGAAGTATGGGTGAGACAGG-3′, and CDK6RTB; 5′-GAAGAAGACTGGCCTAGAG-3′ corresponding to exons 2 and 3 of the mRNA sequence, to give a product of 185 bp. The CDK6 PCR products were transferred to nylon membrane (Hybond-N, Amersham, UK) and hybridized with the oligonucleotide probe CDKI; 5′-GCACTGTAGGCAGATATTC-3′, which is complementary internally to the amplified CDK6 sequence. Primer sequences and conditions used in this work and not described here are available upon request from the authors.
104 or 105 cells were lysed, separated by SDS – PAGE, and probed with rabbit polyclonal anti-CDK6 antibody (Santa Cruz Biotechnology sc-177) using standard protocols (Szepesi et al., 1994). Equivalent protein loading and transfer was verified prior to probing by staining and subsequent destaining with Instaview Nitrocellulose (Merck Ltd, UK).
Lymphocytes were fixed and permeabilized using Facslyse (Becton Dickinson, San José, CA, USA) for 10 min, followed by methanol (30 min at −70°C). Cells were incubated with an anti-CDK6 rabbit polyclonal antibody (Santa Cruz, sc-177), 5 μl in a dilution of 1 : 50, at room temperature for 30 min. Labeling could be blocked by prior addition of a specific blocking peptide (Santa Cruz, sc-177P), in the same dilution. The secondary antibody was FITC labeled and surface labeling was with an anti-CD19-PE mouse monoclonal (Dako). A minimum of 5×103 cells were acquired in the Cellquest program of a FACSCaliber cell sorter (Becton Dickinson). No patient material was available for expression analysis from P3 and P4.
Bullrich F, MacLachlan TK, Sang N, Druck T, Veronese ML, Allen SL, Chiorazzi N, Koff A, Heubner K and Croce CM. . 1995 Cancer Res. 55: 1199–1205.
Chesi M, Nardini E, Brents LA, Schrock E, Ried T, Kuehl WM and Bergsagel PL. . 1997 Nature Genet. 16: 260–264.
Chilosi M, Doglioni C, Yan Z, Lestani M, Menestrina F, Sorio C, Benedetti A, Vinante F, Pizzolo G and Inghirami G. . 1998 Am. J. Pathol. 152: 209–217.
Corcoran MM, Rasool O, Liu Y, Iyengar A, Grander D, Ibbotson RE, Merup M, Wu X, Brodyansky V, Gardiner AC, Juliusson G, Chapman RM, Ivanova G, Tiller M, Gahrton G, Yankovsky N, Zabarovsky E, Oscier DG and Einhorn S. . 1998 Blood 91: 1382–1390.
Costello JF, Plass C, Arap W, Chapman VM, Held WA, Berger MS, Su Huang HJ and Cavenee WK. . 1997 Cancer Res. 57: 1250–1254.
Dascalescu CM, Peoch M, Callanan M, Jacob MC, Sotto MF, Gressin R, Sotto JJ and Leroux D. . 1999 Cancer Genet. Cytogenet. 109: 21–28.
Easton J, Wei T, Lahti JM and Kidd VJ. . 1998 Cancer Res. 58: 2624–2632.
Fischer K, Frohling S, Scherer SW, McAllister Brown J, Scholl C, Stilgenbauer S, Tsui LC, Lichter P and Dohner H. . 1997 Blood 89: 2036–2041.
Goossens T, Klein U and Kuppers R. . 1998 Proc. Natl. Acad. Sci. USA 95: 2463–2468.
Heng HH, Shi XM, Scherer SW, Andrulis IL and Tsui LC. . 1994 Cytogenet. Cell. Genet. 66: 135–138.
Hernandez JM, Schoenmakers EFPM, Cin PD, Michaux L, Van de Ven WJM and Van den Berghe H. . 1997a Genes, Chrom. Cancer. 18: 147–150.
Hernandez JM, Mecucci C, Michaux L, Criel A, Stul M, Meeus P, Wlodarska I, Van Orshoven A, Cassiman JJ, De Wolf-Peeters C and Van den Berghe H. . 1997b Cancer Genet. Cytogenet. 93: 147–151.
Hiom K, Melek M and Gellert M. . 1998 Cell 94: 463–470.
Jaffe EF. . 1997 Am. J. Surg. Pathol. 21: 109–113.
Kaldis P, Russo AA, Chou HS, Pavletich NP and Solomon MJ. . 1998 Mol. Cell. Biol. 9: 2545–2560.
Konig A, Schwartz GK, Mohammad RM, Al-Katib A and Gabrilove JL. . 1997 Blood 90: 4307–4312.
Korsmeyer SJ. . 1992 Annu. Rev. Immunol. 10: 785–807.
Kumar S, White DL, Takai S, Turczynowicz S, Juttner CA and Hughes TP. . 1995 Hum. Genet. 95: 641–644.
Kunz J, Scherer SW, Klawitz I, Soder S, Du YZ, Speich N, Kalff-Suske M, Heng HH, Tsui LC and Grzeschik KH. . 1994 Genomics 22: 439–448.
Latham KM, Eastman SW, Wong A and Hinds PW. . 1996 Mol. Cell. Biol. 16: 4445–4455.
Lorenz W, Schable KF, Thiebe R, Stavnezer J and Zachau HG. . 1988 Mol. Immunol. 25: 479–484.
Mishmar D, Rahat A, Scherer SW, Nyakatura G, Hinzmann B, Kohwi Y, Mandel-Gutfroind Y, Lee JR, Drescher B, Sas DE, Margalit H, Platzer M, Weiss A, Tsui LC, Rosenthal A and Kerem B. . 1998 Proc. Natl. Acad. Sci. USA 95: 8141–8146.
Offit K, Louie DC, Parsa NZ, Noy A and Chaganti RSK. . 1995 Blood 86: 2365–2370.
Oscier DG, Matutes E, Gardiner A, Glide S, Mould S, Brito-Babapulle V, Ellis J and Catovsky D. . 1993 Br. J Haematol. 85: 487–491.
Oscier DG, Gardiner A and Mould S. . 1996 Cancer Genet. Cytogenet. 92: 24–27.
Rabbitts TH. . 1994 Nature 372: 143–147.
Schneider E, Montenarh M and Wagner P. . 1998 Oncogene 17: 2733–2741.
Szepesi A, Gelfand EW and Lucas JJ. . 1994 Blood 84: 3413–3421.
Tambini CE, George AM, Rommens JM, Tsui LC, Scherer SW and Thacker J. . 1997 Genomics 41: 84–92.
Timmermann S, Hinds PW and Munger K. . 1997 Cell Growth Differ. 8: 361–370.
Wahbi K, Hayette S, Callanan M, Gadoux M, Charrin C, Magaud JP and Rimokh R. . 1997 Leukemia 11: 1214–1219.
Wong KF, Chu YC and Hui PK. . 1998 Cancer Genet. Cytogenet. 102: 145–147.
Zeng WR, Scherer SW, Koutsilieris M, Huizenga JJ, Filteau F, Tsui LC and Nepveu A. . 1997 Oncogene 14: 2355–2365.
We thank Anton Kruger for patient sample P2, Samantha Johnson, Sharron Glide, Delin Zhu and Mary Tiller for technical assistance; Anne Gardiner for helpful discussions and critical reading of the manuscript. This work was supported by grants from the Leukemia Research Fund UK and the Medical Research Council of Canada (MRC) to SW Scherer and L-CT Sui. SW Scherer is a Scholar and L-CT Sui a Senior Scientist of the MRC and both are members of the Canadian Genetic Diseases Network.
About this article
Cite this article
Corcoran, M., Mould, S., Orchard, J. et al. Dysregulation of cyclin dependent kinase 6 expression in splenic marginal zone lymphoma through chromosome 7q translocations. Oncogene 18, 6271–6277 (1999). https://doi.org/10.1038/sj.onc.1203033
- 7q translocations
Clinical and biological features of B‐cell neoplasms with CDK6 translocations: an association with a subgroup of splenic marginal zone lymphomas displaying frequent CD5 expression, prolymphocytic cells, and TP53 abnormalities
British Journal of Haematology (2021)
International Journal of Molecular Sciences (2020)
Clinical Case Reports (2020)
CBL-MZ is not a single biological entity: evidence from genomic analysis and prolonged clinical follow-up
Blood Advances (2018)