Article Text

other Versions

Download PDFPDF
Thrombocytosis and STAT5 activation in chronic myelogenous leukaemia are not associated with JAK2 V617F or calreticulin mutations
  1. Samir K Turakhia,
  2. Gurunathan Murugesan,
  3. Claudiu V Cotta,
  4. Karl S Theil
  1. Department of Laboratory Medicine, Robert J. Tomsich Pathology and Laboratory Medicine Institute, Cleveland Clinic, Cleveland, Ohio, USA
  1. Correspondence to Dr Claudiu V Cotta, Cleveland Clinic, 9500 Euclid Avenue/L-30, Cleveland, OH 44195 USA; cottac{at}


Aims Marked thrombocytosis is uncommon in chronic myelogenous leukaemia (CML) but may be associated with mutation of JAK2 V617F, calreticulin (CALR) and/or phospho-STAT5 (p-STAT5) activation in other myeloproliferative neoplasms (MPNs), particularly essential thrombocythaemia (ET). We investigated the JAK2 V617F, CALR and STAT5 activation status in patients with CML and thrombocytosis (CML-T) that mimicked ET, trying to identify a common mechanism for thrombocytosis in MPN.

Methods Blood and bone marrow morphological findings were reviewed from seven CML-T, four otherwise typical CML and one CML in blast phase. All cases were analysed for BCR-ABL1, JAK2 V617F and CALR exon 9 mutation and p-STAT5 expression.

Results Four of seven cases of CML-T had marked thrombocytosis (>1000×109/L). Eleven of 12 cases had megakaryocyte morphology typical for CML. All cases were BCR-ABL1 positive. Eleven of 12 cases were negative for JAK2 V617F, while STAT5 was activated in six of seven CML-T and in four of five CML cases. No case had a detectable CALR exon 9 mutation. One case of CML developed ET-like morphology and had JAK2 V617F detected while in molecular remission for CML.

Conclusions Detection of BCR-ABL1 is critical in the distinction of ET from CML. Thrombocytosis and STAT5 activation in CML-T are not consistently associated with CALR exon 9 or JAK2 V617F mutation.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Chronic myelogenous leukaemia (CML) is a myeloproliferative neoplasm (MPN) defined by the presence of a BCR-ABL1 fusion, which leads to the expression of an oncoprotein with constitutive tyrosine kinase activity.1 While CML classically presents with a neutrophilic leucocytosis with left shift, eosinophilia and basophilia, these features are neither pathognomonic nor found in every case. Similarly, thrombocytosis, defined as >400×109 platelets/L, may be seen in up to 30% of patients with CML, but is also common in other MPNs including polycythaemia vera (PV), essential thrombocythaemia (ET) and the prefibrotic phase of primary myelofibrosis (PMF). A marked thrombocytosis, defined here as a platelet count >1000×109/L, is rare in CML at presentation, however.2 Among MPNs, marked thrombocytosis, especially in the absence of prominent leucocytosis, is typically associated with ET. Some cases of ET can mimic CML with thrombocytosis (CML-T), and a final diagnosis of CML may come as a surprise when conventional cytogenetic and/or molecular studies reveal an unanticipated t(9;22)(q34;q11.2) and/or BCR-ABL1 fusion.3 The concept of Philadelphia chromosome-positive (Ph+) or BCR-ABL1-positive ET has been debated in the literature,4–6 but the World Health Organization Classification of Tumours: Pathology and Genetics of Tumours of Hematopoietic and Lymphoid Tissues resolved the issue in 2001 by establishing BCR-ABL1 as both a disease-defining criterion for the diagnosis of CML and an exclusionary criterion for the diagnosis of ET, PV and PMF.7 The availability of targeted tyrosine kinase inhibitor (TKI) therapy against BCR-ABL1 makes identification of this fusion transcript of critical importance.3

In contrast to other MPNs, the molecular mechanisms responsible for the marked thrombocytosis seen in some cases of CML have not been identified. Because it has been shown that up to 2.5% of patients with CML may harbour a JAK2 V617F mutation, screening for JAK2 mutations in patients with CML that develop thrombocytosis, either at presentation, or during or after treatment is recommended.8–10 Still, it is not clear whether this mutation is present in all cases of CML-T, nor whether all cases of CML with JAK2 mutations manifest with thrombocytosis. To further complicate the issue, in addition to JAK2, other genes, including MPL and calreticulin (CALR), can be mutated in MPN with thrombocytosis, implicating activation of downstream transcription regulation pathways as a potential mechanism. In particular, activation of STAT5 transcription regulators resulting in a change in localisation of phospho-STAT5 (p-STAT5) from the cytoplasm to the nucleus has been demonstrated in CML, ET and PMF. Immunohistochemistry (IHC) of bone marrow core biopsies can be used to demonstrate this aberrant nuclear localisation of p-STAT5 in megakaryocyte nuclei in cases of ET and PMF. In these entities, unlike the cytoplasmic staining seen in normal megakaryocytes, the abnormal megakaryocytes display nuclear p-STAT5 similar to normal erythroid progenitors providing morphological evidence of in vivo JAK2 activation and correlating with the presence of JAK2 V617F or MPL W515L.11 ,12

In this study, we identified six cases of CML-T at presentation and one case of CML that developed thrombocytosis during TKI therapy, and compared clinical, morphological and cytogenetic features with four cases of otherwise typical CML and one case of CML in megakaryoblastic crisis. All cases were evaluated for JAK2 V617F or CALR exon 9 mutations, and STAT5 activation status was investigated by IHC.

Materials and methods

Case selection

After Internal Review Board approval, we identified cases diagnosed as CML that presented with thrombocytosis at the time of bone marrow evaluation during 2009–2014. The diagnosis of CML was confirmed by demonstrating the Philadelphia chromosome (Ph) by metaphase cytogenetics and/or BCR-ABL1 fusion by interphase fluorescence in situ hybridisation (FISH). Clinical history, histopathological and treatment data were reviewed and collated, along with JAK2 mutation studies, when available. As a comparison group, four cases of CML with an otherwise typical presentation and one case of CML in megakaryoblastic crisis were reviewed.


Immunohistochemical stains for p-STAT5 were performed on 4 μm thick sections of formalin-fixed paraffin-embedded (FFPE) bone marrow biopsy cores from each case using an automated stainer (Benchmark XT or Benchmark Ultra, Ventana Medical Systems (VMS), Tucson, Arizona, USA). To detect the protein–antibody complex, a biotin/streptavidin–horseradish peroxidase/diaminobenzidine tetrahydrochloride (DAB) detection kit (iView DAB Detection; VMS; Cat. No. 760-091) was used. The antigen retrieval conditions for p-STAT5a/b mouse monoclonal clone Y694/99 (Advantex BioReagents, Houston, Texas, USA; Cat. No. AX1) were as follows: Cell Conditioning one buffer (VMS; Cat. No. 950-124) at 95°C for 20 min, 1:50 dilution, incubate for 60 min at room temperature, followed by amplification with iView DAB Detection kit.

p-STAT5 staining was considered positive if >10% of the megakaryocytes showed weak or strong nuclear staining at an intensity greater than the background. Staining quality was evaluated in the context of appropriate internal positive control (erythroid, megakaryocyte cytoplasm) and negative control (granulocytes) cells.


Cells from unstimulated peripheral blood or bone marrow aspirates were cultured for 24–48 h in Gibco MarrowMAX medium (Life Technologies, Grand Island, New York, USA; No. 12260-014) with or without added granulocyte-macrophage colony-stimulating factor. Cells were harvested using an automated harvester (HANABI-PI, Adstec, Funabashi, Chiba, Japan). After harvesting, the cell pellet was dropped onto slides in a humidified drying chamber (CDS-5, Thermotron, Holland, Michigan, USA). After heat treatment GTG-banding was performed using an automated platform (Little Dipper Processor, SciGene, Sunnyvale, California, USA). Slides were scanned and metaphase cells analysed using an automated scanner and image analysis system (GenASIs, Applied Spectral Imaging, Carlsbad, California; USA). At least 20 metaphase cells were analysed wherever possible, and karyotypes were described according to ISCN (2013).13


Slides were dried at room temperature overnight, followed by immersion in 73°C 2× saline-sodium citrate (Fischer Scientific, Pittsburgh, Pennsylvania, USA) for 2 min. Slides were then incubated for 10 min in 37°C protease solution (Abbott Molecular, Des Plaines, Illinois, USA; No. 02J03-032). Slides were washed with phosphate-buffered saline, fixed in formaldehyde, and dehydrated with alcohol. Ten microlitres of a probe mixture consisting of 7 μL LSI hybridisation buffer (Abbott Molecular, No. 06J67-011), 2 μL purified water and 1 μL LSI BCR SpectrumGreen/LSI ABL SpectrumOrange dual colour, single fusion probe (Abbott Molecular, No. 05J77-001) was added to each slide. The slides were sealed and heated to 73°C for 5 min. Slides were held at 37°C overnight for hybridisation. The coverslips were removed and slides were incubated in 0.4× SSC/0.3F NP-40. They were then counterstained with Vectashield with 4′6-diamidino-2-phenylindole mounting medium (Vector Laboratories, Burlingame, California, USA; Cat. No. H-1200) and coverslipped. Signals from 200 non-overlapping nuclei were counted using an Olympus BX40 fluorescence microscope (Olympus, Tokyo, Japan). Cases showing >10% of cells with fusion signals were reported as positive for BCR-ABL1.

JAK2 V617F mutation analysis

DNA was extracted from either peripheral blood or bone marrow aspirate samples at the time of diagnosis, or from FFPE bone marrow aspirate clot samples. After DNA extraction was complete, a mixture of 5 μL LightCycler TaqMan Master (Roche, Indianapolis, Indiana, USA; Cat. No. 04535286001), 2.5 μL IPSOGEN primers and probes (Qiagen, Valencia, California, USA; Cat. No. 673113) and 12.5 μL nuclease-free water was added to the extracted DNA. The sample was run on a LightCycler 480 Real-Time PCR system (Roche; Cat. No. 05015278001). If the mean ratio of FAM/VIC of a case was less than that of the 2% of mutant allele control, the sample was designated wild type. If the ratio was greater than that of the 50% mutant allele control, it was designated homozygous for the JAK2 V617F mutation. Ratios between 2% and 50% were designated heterozygous for the JAK2 V617F mutation.

CALR mutation analysis

DNA previously extracted from peripheral blood or bone marrow aspirate samples was used, or if not available, DNA was extracted from FFPE bone marrow clot sections. CALR gene fragment analysis was performed using the method and primers previously described.14 Briefly, CALR exon 9 target sequences from genomic DNA (gDNA) were generated using a touch-down PCR programme. The PCR reaction contained 10 μL of HotStarTaq Master Mix (2x; Qiagen), 0.6 μL of 2 µM primer pair (FAM-labelled forward primer and unlabelled reverse primer; final primer concentration 60 nM), 1 μL containing 20 ng of gDNA, and 8.4 μL water to a total volume of 20 μL. PCR fragment analysis was performed by sizing fluorescently labelled PCR products on a ABI 3730 DNA Analyzer (Applied Biosystems, Foster City, California, USA), and results analysed using Gene Mapper V.4.1 software.


Clinical and cytogenetic findings

A total of six cases of CML presenting with thrombocytosis (CML-T) at the time of initial bone marrow evaluation were identified, along with one case that developed thrombocytosis during the course of imatinib therapy. These were compared with four cases of CML with an otherwise typical presentation and one case of CML in megakaryoblastic crisis. The clinical and cytogenetic findings are summarised in table 1. The diagnosis of CML was confirmed in all cases by demonstration of t(9;22)(q34;q11.2) by metaphase cytogenetics and/or FISH for BCR-ABL1. In all cases FISH for BCR-ABL1 was performed, in 10 of 12 cases on the peripheral blood and in 2 cases on the bone marrow aspirate. Three of four cases with platelet counts >1000×109/L showed a mixture of karyotypically normal and Ph+ cells, whereas only one of the remaining cases, in which cytogenetic studies were performed after initiation of TKI therapy, showed any normal cells. None of the four cases with platelet counts >1000×109/L had splenomegaly. Thrombotic events were not reported in any of the patients in the study. While no patient with CML-T had bleeding events, one patient with typical CML reported epistaxis with an onset significantly before the onset of the myeloid neoplasm.

Table 1

Clinical and cytogenetic findings

Complete blood count and peripheral blood morphology

Cases of CML-T at presentation (table 2, cases 1–6) had platelet counts ranging from 667 to 1990×109/L; in 4/6 (67%) the platelet count exceeded 1000×109/L. Interestingly, of the cases with marked thrombocytosis, one case presented solely with thrombocytosis, one with concurrent anaemia, one with a very mild leucocytosis and one with concurrent anaemia and leucocytosis. Giant platelets were noted in two of the cases with marked thrombocytosis. The two cases (cases 5 and 6) that presented with thrombocytosis <1000×109/L had markedly elevated white cell counts. Apart from the case of CML on TKI therapy (case 7), all but one (case 2) showed a granulocytic left shift in the peripheral blood, irrespective of the platelet count. The cases of otherwise typical CML in chronic phase had a leucocytosis with or without anaemia (cases 8–11) and, in blast phase, had leucocytosis, anaemia and thrombocytopenia (case 12).

Table 2

Complete blood count data and peripheral blood morphology

Bone marrow morphology

Bone marrows were hypercellular in 6/7 patients with CML-T and in 5/5 in the comparison group. Megakaryocytes were increased in all seven cases of CML-T; in 4/7 cases megakaryocytes were clustered (cases 1, 4, 5 and 7). In the comparison group, 2/5 showed increased megakaryocytes without clustering (cases 9 and 12) and the remaining 3 cases showed a normal number of megakaryocytes (cases 8, 10 and 11). In 11/12 total cases megakaryocytes were small with hypolobate nuclei characteristic of CML (figure 1; table 3). In the remaining case (case 7) the megakaryocyte morphology resembled that seen in typical ET with large forms with complex lobations. Interestingly, the latter case was the only one with a JAK2 V617F mutation. Stromal fibrosis was assessed using trichrome and/or reticulin stains. The case with JAK2 V617F mutation showed reticulin fibrosis; this finding was not present at the time of the initial diagnosis and became evident only in specimens obtained after the initiation of the TKI therapy when the overall bone marrow findings corresponded to a non-CML MPN.

Table 3

Bone marrow morphology, p-STAT5 IHC and molecular data

Figure 1

Bone marrow histopathology in chronic myelogenous leukaemia (CML) with marked thrombocytosis. (A and B) Bone marrow aspirates from case 1 and case 4, respectively, show increased numbers of small megakaryocytes, many with hypolobate nuclei. (C and D) Bone marrow biopsies from case 1 and case 4, respectively, show marked megakaryocytic hyperplasia with clustering and hypolobate nuclei. (A and B) Wright–Giemsa, ×260; (C and D) H&E, ×260.

IHC for p-STAT5

Ten of the 12 cases showed abnormal localisation of p-STAT5 to megakaryocyte nuclei (figure 2A). In two cases (cases 6 and 12) p-STAT5 staining was weakly positive in <10% of megakaryocyte nuclei and interpreted as negative (figure 2B). All cases with a platelet count >1000×109/L were positive for p-STAT5.

Figure 2

Immunohistochemistry for phospho-STAT5 (p-STAT5). (A) Bone marrow biopsy from case 6 shows normal cytoplasmic and absent nuclear localisation of p-STAT5. (B) Bone marrow biopsy from case 5 shows abnormal nuclear localisation of p-STAT5 in most megakaryocyte nuclei. (A and B) Immunohistochemistry for p-STAT5, ×260.

JAK2 V617F mutation analysis

JAK2 V617F mutation analysis was informative in all cases (table 3). All were negative for the mutation except for case 7: in this case thrombocytosis and anaemia persisted while on imatinib and subsequently resolved following initiation of anti-JAK2 therapy.

CALR mutation analysis

Molecular analysis to detect mutations in CALR exon 9 corresponding to either the 5 bp insertion (267 bp fragment) or 52 bp deletion (210 bp fragment) was attempted in all 12 cases (table 3). Eight cases were informative and yielded the wild-type fragment (262 bp). In four cases (cases 1, 6, 7 and 11) insufficient DNA was amplified; these were considered assay failures due to either limited amount or suboptimal DNA integrity.

Treatment course and follow-up

All patients received TKI therapy aside from case 1, for whom treatment and follow-up data are not known (table 4). Of the three patients with marked thrombocytosis for whom follow-up data was available, two showed normalisation of the platelet count with TKI therapy, despite absence of molecular remission (MR) (table 4). Two of three patients with lesser degrees of thrombocytosis showed normalisation of their platelet counts after TKI therapy and achieved an MR. The third patient (case 7) did not achieve MR and experienced adverse side effects while on TKI resulting in medication compliance issues; the platelet count normalised after subsequent identification of JAK2 V617F mutation and initiation of anti-JAK2 therapy. In the comparison group, four cases of CML without thrombocytosis (cases 8–11) achieved MR. The remaining case of CML in blast crisis (megakaryoblastic) was treated with multiple TKIs and an umbilical cord blood transplant, but relapsed after transplant and nilotinib therapy, and eventually expired due to fungaemia.

Table 4

Treatment and clinical follow-up


Thrombocytosis is unusual in CML at presentation, especially when it is the only abnormality in the peripheral blood. Prior to widespread adoption of the WHO classification, such cases were often classified as Ph+ or BCR-ABL1-positive ET, and were sufficiently uncommon to merit case reports in the literature. A review of 3 new and 20 reported cases described as Ph+ ET noted that the combination of peripheral blood thrombocytosis with normal leucocyte count and haemoglobin level, and small, round monolobate megakaryocytes in the marrow was a diagnostic clue for recognising Ph+ thrombocythaemia as the presenting feature of CML.15 These cases showed a female predominance, presented with bleeding or thrombotic complications, and typically lacked splenomegaly. Marrow megakaryocytes were increased, and the small megakaryocytes were clearly different from the enlarged hyperlobate megakaryocytes associated with typical ET (Ph). Some patients eventually developed a typical peripheral blood picture of CML within 2–3 years of follow-up, with a high risk of disease progression and blastic transformation. Because this clinical course more closely resembled CML rather than ET, so-called Ph+ ET came to be considered a clinical variant of CML.

We identified seven cases of CML-T at the time of bone marrow evaluation; the clinical and morphological findings in four cases with marked thrombocytosis are remarkably similar to those described by Michiels et al.15 Thrombocytosis was detected at presentation in six patients, while in one the increased platelet count developed during TKI therapy. Leucocytosis was a feature in 4/7 cases, while all cases showed a granulocytic left shift in the peripheral blood. Bone marrow biopsies showed small, hypolobated megakaryocytes characteristic of CML in 6/7 cases. There was only one exception: in the JAK2 V617F+ case (case 7) megakaryocyte morphology resembled that seen in typical ET and the thrombocytosis persisted during TKI treatment. Treatment with TKI resulted in normalisation of the platelet counts in all cases, except case 7 that was positive for JAK2 V617F.

While cases of CML presenting with thrombocytosis (CML-T) show phenotypic variability ranging from ET-like to otherwise typical CML, little is known about the mechanism(s) responsible for the elevated platelet count and how it may differ from that seen in other MPNs. To investigate possible mechanisms responsible for the elevated platelet count in our cases, DNA was extracted from bone marrow and tested for mutations involving JAK2 codon 617 and CALR exon 9; we did not examine mutations involving JAK2 exons 12–15 or MPL in this study, mainly because the amount of DNA we were able to extract from the diagnostic specimens was limited. The JAK2 V617F mutation was present in only one case and none showed mutation of CALR. Thus, although our study is small, the absence of JAK2 and CALR mutations suggests that they do not play a role in causing the thrombocytosis in most cases of CML-T.

In JAK2+ MPN, activation of STAT5 is a major underlying mechanism for uncontrolled proliferation in the megakaryocytic lineage. Upon activation through phosphorylation, this transcription regulator changes its localisation from the cytoplasm to the nucleus, thereby leading to changes in gene expression patterns. Abnormal nuclear localisation of p-STAT5 in megakaryocytes, as detected by IHC, has been found in most MPNs in association with mutated JAK211 or MPL.12 The majority (6/7) of our CML-T cases were positive for p-STAT5 by IHC, while only one case was positive for the JAK2 V617F mutation, indicating that activation of this transcription regulation pathway can occur by other mechanisms. Interestingly, most cases of CML without thrombocytosis included in this study as controls also showed p-STAT5 activation by IHC, despite absence of JAK2 V617F mutation.

A possible explanation for the abnormal nuclear localisation of p-STAT5 in CML is the biological effect of BCR-ABL1. It has been shown that the BCR-ABL1 oncoprotein interacts with other cytoplasmic molecules to dysregulate multiple cellular processes, including the JAK-STAT pathway, leading to transcription of genes for cell survival and proliferation.1 ,11 Through a JAK-independent manner, the BCR-ABL1 oncoprotein leads to constant activation through phosphorylation of STAT5 by direct association of STAT SH2 domains and phosphorylated tyrosines on BCR-ABL1.1 In the nucleus, p-STAT5 binds to DNA, leading to transcription of genes necessary for cell survival and proliferation,11 probably through the same mechanism described in other MPNs. It is unclear whether mutated JAK2 or BCR-ABL1 proteins are equally potent activators of STAT5. Interestingly, activation of either pathway must not consistently result in megakaryocyte proliferation, since not all JAK2+ MPNs present with thrombocytosis nor do most CML. In these circumstances, STAT5 activation in megakaryocytes must be achieved through other mechanisms. Since both JAK2 and BCR-ABL1 mutations are thought to be present at a stem-cell level, there is a distinct possibility that they could induce the production of cytokines by other bone marrow components and that these cytokines may be at least partially responsible for megakaryocytic proliferation and thrombocytosis.

Careful attention to bone marrow histopathology in MPN can reveal differences in megakaryocyte morphology that help discriminate between diagnostic entities.5 ,15–19 For example, the megakaryocytes and their nuclei are larger in ET as compared with other MPNs,18 whereas the megakaryocytes in CML are smaller than normal and hypolobated.20 Since STAT5 is usually activated in all these entities, the classic ‘dwarf’ megakaryocyte morphology in CML cannot be attributed to this factor alone. Other, as yet unknown, mechanisms, possibly related to BCR-ABL1 function, may be responsible for the aberrant megakaryocyte morphology in CML.

In most cases of CML, treatment with TKI usually leads to normalisation of bone marrow morphology and platelet count.21 This was the outcome in 4/6 cases of CML-T for which follow-up data was available in our study. One case (case 3) had persistence of thrombocytosis after 6 months of TKI therapy, but then was lost to follow-up. The notable exception was the case of CML that was subsequently found to be positive for JAK2 V617F: the platelet count normalised only after anti-JAK2 therapy (ruxolitinib) was initiated. This patient was initially diagnosed with CML and achieved MR for 3 months, but was found to harbour a JAK2 V617F mutation when testing was performed to evaluate persistent thrombocytosis. We were unable to retrospectively determine whether the JAK2 V617F mutation was present at diagnosis of CML. Comparison with similar cases in the literature10 ,22 ,23 suggests two possible scenarios: either both mutations occur sequentially in the same progenitor cell (JAK2 followed by BCR-ABL1) with disease manifesting as CML,24 or the mutations occur independently in two distinct clones. In either situation, TKI therapy would be expected to suppress growth of BCR-ABL1+ cells with re-emergence of the JAK2+ clone. Available evidence in the literature favours the sequential model.25

In conclusion, this study validates the importance of determining BCR-ABL1 status in cases of marked thrombocytosis clinically suspected to be ET.3 An accurate diagnosis of CML can help avoid unnecessary antithrombotic therapy and allow rapid initiation of effective TKI therapy. STAT5 activation through BCR-ABL1 probably contributes to the thrombocytosis seen in a subset of cases. The mechanisms resulting in this activation are not completely identical to those described in JAK2 or CALR mutation-positive MPN and can be reversed by TKI therapy. Thrombocytosis and STAT5 activation occurring in CML-T are not consistently associated with JAK2 or CALR mutations.

Take home messages

  • Chronic myelogenous leukaemia BCR-ABL1 positive can have an initial presentation mimicking essential thrombocythaemia.

  • In cases of CML with thrombocytosis the morphology of the megakaryocytes is similar to that of megakaryocytes in CML without thrombocytosis.

  • Investigation of BCR-ABL1 status is necessary in the de novo diagnosis of myeloproliferative neoplasms.



  • CVC and KST contributed equally.

  • This work is based on an abstract presented in poster form at the 104th Annual Meeting of the United States/Canadian Academy of Pathology, Boston, MA, 25 March 2015.

  • Handling editor Mary Frances McMullin

  • Contributors SKT, GM, CVC and KST conducted research. SKT, CVC and KST wrote the article. CVC and KST designed the study.

  • Competing interests None declared.

  • Ethics approval Cleveland Clinic Internal Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.