Sporadic and Familial Acute Myeloid Leukemia with CEBPA Mutations

Purpose of Review CCAAT enhancer binding protein A (CEBPA) gene mutation is one of the common genetic alterations in acute myeloid leukemia (AML), which can be associated with sporadic and familial AML. Recent Findings Due to the recent advances in molecular testing and the prognostic role of CEBPA mutation in AML, the definition for AML with CEBPA mutation (AML-CEBPA) has significantly changed. This review provides the rationale for the updates on classifications, and the impacts on laboratory evaluation and clinical management for sporadic and familial AML-CEBPA patients. In addition, minimal residual disease assessment post therapy to stratify disease risk and stem cell transplant in selected AML-CEBPA patients are discussed. Summary Taken together, the recent progresses have shifted the definition, identification, and management of patients with AML-CEBPA.

Historically, in intermediate cytogenetic risk AML, CEBPAdm mutations have been associated with favorable prognosis in contrast to CEBPA-wild type (WT) or CEB-PAsm, particularly those harboring the typical CEBPAdm pattern of concurrent N-terminal and C-terminal/basic DNA binding and leucine zipper (bZIP) domain mutations [10,11,15,17]. Several recent studies have demonstrated similarly favorable outcome in bZIP-mutated cases irrespective their double or single CEBPA-mutated status, particularly with bZIP in-frame mutations [12][13][14]. These findings refine the landscape of CEBPA in AML risk stratification and have been adopted in the 5 th edition of WHO Classification (WHO-5) and International Consensus Classification (ICC) [19,20]. In this review, we discuss the current knowledge of sporadic and familial AML-CEPBA, including the recent advances, and their impact on the laboratory evaluation and clinical management. We also discuss the treatment strategies for patient with CEBPA mutations, and the role of minimal residual disease (MRD) and stem cell transplant in this patient population.

CEBPA Protein Structure
CEBPA is a member of the basic region leucine zipper family composed of six transcription factors. It is encoded by the intronless CEBPA gene located on chromosome 19q13.1, containing two transactivation domains (TAD1 and TAD2) in the N-terminus regulating transcription activity, and a basic DNA binding (DBD) and leucine zipper (ZIP) bZIP domain in the C-terminus responsible for DNA binding and dimerization with CEBPA protein family members (Fig. 1). By means of two alternative protein translation initiation sites in the N-terminus, CEBPA generates two isoforms, a full-length 42 kDa (p42) and a truncated 30 kDa (p30) (Fig. 1). The ratio of p42 to p30 appears important for myeloid differentiation and excessive p30 isoform has been shown to act as a dominant negative on the remaining p42 isoform, inhibiting the terminal differentiation of granulocytes [16,21,22].

Sporadic AML with CEBPA Mutations
CEBPA mutations are observed in about 5-16% of de novo AML and can be separated into subgroups of CEB-PAdm (50-60%) and CEBPAsm (40-50%). CEBPAdm predominantly show a N-terminal (p.1-120, transcript ID NM_004364.4) frameshift insertion/deletion (indel)-type mutation (less frequently nonsense mutation), along with a concurrent bZIP (p.278-358) in-frame indel (less frequently missense mutation). Early cloning studies have shown that CEBPAdm are frequently biallelic with simultaneous in trans occurrence of the N-and C-terminal/bZIP mutations on distinct alleles [11,17]. A small subset of patients may harbor two CEBPA mutations not conforming to the typical CEBPAdm pattern, e.g. two N-terminal mutations, two C-terminal/bZIP mutations, one middle and one C-terminal/ bZIP mutations. Rarely, homozygous single N-and C-terminal/bZIP mutations have been reported as a result of loss of heterozygosity arising from acquired uniparental disomy of chromosome19q13.1 [23,24]. In CEBPAsm, mutations do not show a clear localizing pattern, and may occur in the N-terminal, C-terminal/bZIP, or middle region of the gene [10,11,15,17]. CEBPAsm-bZIP occurs less frequently than CEBPAsm-nonbZIP in adults [12,13]. Conversely, CEB-PAsm-nonbZIP N-terminal mutations are rare in pediatric AML [3,14]. The N-terminal frameshift/nonsense mutations enforce translation of the aberrant dominant-negative p30 isoform from the second protein translation initiation codon and thus suppress myeloid differentiation, whereas the C-terminal/bZIP in-frame indels or missense mutations disrupt protein dimerization and DNA binding. The combination of N-and C-terminal/bZIP mutations disrupt normal CEBPA function and downstream cellular processes such as cell cycle arrest and myeloid differentiation [16].
Comutation of other genes occurs more frequently in CEBPAsm (up to 90%) than in CEBPAdm, suggesting that gain of secondary mutations may be required for leukemogenesis, whereas a second CEBPA mutation in CEBPAdm may be sufficient to drive leukemogenesis without additional cooperating mutations. In a comprehensive analysis of 244 adult AML-CEBPA, Fasan et al. reported multiple comutations in the CEBPAsm cohort, including FLT3-ITD, NPM1, ASXL1, IDH1/2, and RUNX1 with the majority (70%) of the CEBAPsm mutations located outside of the bZIP region [10]. Conversely, CEBPAdm cases were enriched with mutations  The typical CEBPAdm mutation pattern in sporadic AML is concurrent presence of a N-terminal frameshift/nonsense mutation and a C-terminal/bZIP in-frame indel or missense mutation. In familial CEBPA-mutated AML, the dominant mutation pattern is similar to that of typical CEBPAdm, with a N-terminal germline mutation and a C-terminal somatic mutations. Indel: insertion and/or deletion of transcription factors GATA2 and WT1 and epigenetic modifiers TET2 and ASXL1. FLT3-ITD and NPM1 mutations were far less commonly seen in CEBPAdm [9,11]. Clinically, CEBPAdm patients are associated with younger age, higher hemoglobin levels, higher WBC counts, lower platelet counts, and higher bone marrow blast percentages, compared with CEBPA-wild type (WT) and CEB-PAsm cases [9,10,17,[25][26][27]. Morphologic and phenotypic findings are mostly consistent with AML with or without maturation and a subset of cases show myelomonocytic or monoblastic differentiation without significant difference seen between CEBPAdm and CEBPAsm [10,11,17,28]. The blast phenotype of CEBPAdm is characterized by frequent expression of CD34 and some myeloid associated antigens (CD11b, CD15, CD13, CD33, CD65) with a higher frequency of CD7, CD15, and HLA-DR expression, but a lower frequency of CD56 expression than that of CEBPAsm [10,29,30]. CEBPAdm is strongly associated with a normal karyotype and deletion of chromosome 9 is the most frequent cytogenetic abnormality [10]. Erythroid or multilineage dysplasia is observed in a quarter of cases, though these features do not influence clinical outcomes [31].
Early studies with combined CEBPAsm and CEBPAdm cases revealed that presence of CEBPAm was associated with superior clinical outcomes in AML with a normal karyotype [16,32]. Subsequent data showed that the favorable prognosis with conventional chemotherapy was restricted to CEBPAdm in the absence of concurrent FLT3-ITD in contrast to CEBPAsm or CEBPA-WT [10,11,15,17,29,33]. Therefore, the 2016 WHO revision recognized AML with biallelic mutation of CEBPA as a distinct entity of AML, and AML with germline CEBPA mutation was placed under a separate subcategory of myeloid neoplasms with germline predisposition [34]. Both the European LeukemiaNet (ELN) and National Comprehensive Cancer Network guidelines included biallelic mutated CEBPA in their prognostic classifications as a favorable risk marker [35].
Interestingly and importantly, several recent studies from large-scale adult and pediatric AML cohorts demonstrated similarly favorable clinical outcome in bZIP-mutated AML irrespective their double or single CEBPA-mutated status (CEBPA-bZIP), particularly with in-frame bZIP mutations. CEBPA-bZIP cases also shared similar clinical and mutational characteristics with CEBPAdm [12][13][14]. Given the lower prevalence of CEBPAsm-bZip (1-2% of AML), previous studies may not have been adequately powered to evaluate the prognostic significance of this subset and the recent studies demonstrate the power of large-scale cohort studies in the analysis of clinical, biology, and prognostic impact of AML [12][13][14].
More specifically, Tarlock et al. studied the largest reported pediatric/young adult (< 30 years) AML series (n = 2958) and observed equally favorable clinical outcome in CEBPAdm and CEBPAsm cases harboring a bZIP mutation in comparison to CEBPA-WT patients, including a higher MRD-negative complete remission rate at the end of induction, identical superior 5-year event free survival (EFS) and overall survival (OS), and similarly lower relapse risk [14]. The findings were confirmed in 2 subsequent studies. Taube et al. analyzed 4078 adult AML patients and found similarly favorable OS and EFS in CEBPAdm and CEBPAsm-bZIP, but not CEBPAsm-TAD (with a single N-terminal mutation), in comparison to CEBPA-WT. When further regrouped the patients according to the presence and absence of in-frame bZIP mutations, only patients harboring in-frame bZIP mutations showed better outcome in contrast to other CEBPA mutations [13]. Wakita et al. reported a multicenter joint study of 1028 Japanese AML patients aged ≥ 16 years demonstrating a strong association of favorable prognosis with bZIP mutations and validated CEBPAdm as a cofounding factor overlapping with bZIP mutations. These recent studies also found similar clinical and biological features among CEBPAdm and CEBPAsm-bZIP cases. In the pediatric group, CEBPAdm and CEBPAsm-bZIP patients showed a comparable gene and mRNA expression profile and prevalence of cooperating mutations, including GATA2, CSF3R, NRAS, FLT3-ITD, WT1, and nearly absence of NPM1 mutations. In comparison to CEBPA-WT, CEBPAdm and CEBPAsm-bZIP patients were older; had higher white blood cell counts; more likely to have a normal karyotype; and had a higher incidence of GATA2 or CSF3R mutations, lower incidence of NPM1 mutation, and a similar prevalence of FLT3-ITD and NRAS mutations. In adult cases, CEB-PAdm and CEBPAsm-bZIP are of significantly younger age and have higher rates of CD34-positive blasts compared to CEBPA-WT and CEBPAsm-nonbZIP [12][13][14].
Comutations may impact on the prognosis of CEBPAmutated AML. In adult CEBPAdm, the favorable outcome of CEBPAdm may be lost with a concurrent FLT3-ITD mutation [11,36]. GATA2 and NPM1 comutations were found to be associated with superior OS, though this finding has not been consistently replicated. Similarly, possible negative impact on prognosis were found in association with TET2, DNMT3A, WT1, CSF3R, ASXL1, or KIT comutations by some but all studies [8,10,26,27,[37][38][39][40][41][42][43]. The controversial results are likely attributable to the small sample sizes limiting the analysis power. In childhood CEBPAdm, CSF3R comutation was associated with poor relapse-free survival, but OS was not significantly different [38]. Distinct from adult AML, FLT3-ITD occurs with similar frequency in both CEBPAsm and CEBPAdm, and showed no effect on OS or EFS [6,14].
In the recent large cohort studies, GATA2 mutations were found to be enriched in both pediatric and adult patients with CEBPAdm and CEBPAsm and did not show clear prognostic significance in some studies [13,14] while one study showed that a GATA2-mutated/WT1-wild type comutation signature carried a significantly better OS in adult patients with CEBPA-bZIP and an intermediate-risk karyotype [12]. GATA2 and CSF3R, as well as GATA2 and WT1, are mutually exclusive in CEBPA-mutated AML (Table 1). WT1 mutations are frequently detected in adult CEBPAdm and CEBPAsm-bZIP, whereas CSF3R mutations are rare (~ 3%) in adults but are more common in children (13%) [12][13][14]. In pediatric study with CEBPA-bZIP, concomitant CSF3R mutations exhibited a lower EFS and a higher relapse rate with no impact on OS, confirming findings from previous studies [14,38].

Familial AML with Germline CEBPA Mutation
It is well recognized that myeloid neoplasms may occur in associated with inherited or de novo germline mutations. The diagnostic category of myeloid neoplasms with germline predisposition was added in the 2016 WHO classification given their increasing recognized prevalence and importance in clinical management of the patients and families [44]. Although they are rarer as compared to the sporadic myeloid neoplasms, the list likely will expand as their recognition and detection grow.
In FAML-CEBPA, the CEBPA mutation pattern is mainly reminiscent of that of the sporadic CEBPAdm. The vast majority show a typical CEBPAdm pattern with a heterozygous frameshift germline mutation clustering in the N-terminus and a concurrent C-terminal/bZIP in-frame indel somatic mutation at the time of AML diagnosis. Cases with N-terminal germline mutation show near complete penetrance (90%) of AML development with typical onset at a median age of 25 years (range: 1.75-46) from a report of 10 families with 24 affected individuals [18]. Interestingly, the bZIP somatic mutations are not stable and disease relapse frequently shows a novel leukemic clone with somatic mutations distinct from the diagnostic sample, e.g., a different bZIP mutation, a second N-terminal mutation, or loss of the somatic CEBPA mutation [18].
Additionally, rarer FAML-CEBPA cases with germline CEBPA mutations located outside the N-terminal region have been described, involving the C-terminal/bZIP or middle region. They interestingly exhibited incomplete penetrance in contrast to the high penetrance observed in the N-terminal germline cases. In a 45-year follow-up of a large family pedigree, Pathak et al. observed a germline bZIP missense mutation p.Q311P showing an incomplete AML penetrance of 46% in 6 of 13 confirmed mutation carriers or obligate carriers. Despite the findings of attenuated DNA-binding and transactivation activities of the p.Q311P mutant in in vitro assays, seven other confirmed mutation carriers showed no evidence of hematologic malignancy with an age range of 24 to 88 years [48]. The lower penetrance of C-terminal germline mutations was also confirmed by other reports [51,59,63]. Additionally, Molecular profile at relapse Relapse and evolution of the diagnostic clone New clone distinct from the diagnostic clone three individuals were described to have germline bZIP missense mutations without a positive family history [9]. Two other groups reported a middle-region germline mutation p.E148* in 2 FAML-CEBPA patients with confirmed mutation carriers in the families free of hematologic malignancy at ages 37, 59, and 66. Both index patients harbored a concurrent somatic N-terminal frameshift or nonsense mutation [60,64]. Mendoza et al. summarized the published FAML-CEBPA cases and compared the age of AML onset between germline mutations of different locations [64]. FAML-CEBPA with N-terminal germline mutations occur with the greatest frequency between ages 21 and 30 years and typically develops before age 50, whereas C-terminal germline mutations occur later with the greatest frequency between ages 51 and 60 years. The reduced penetrance and late onset of FAML-CEBPA patients with germline C-terminal or middle-region mutations complicate the clinical recognition of these cases and likely contributed to its apparent rarity and probable underestimation. Therefore, the persistence of a CEBPA mutation at the time of complete remission warrants further germline testing, irrespective of family history. Based upon the cohort with N-terminal germline mutations, the clinicopathologic and molecular characteristics of FAML-CEBPA were found to resemble those of its sporadic counterpart [18]. They are commonly associated with FAB subtypes M1 and M2, with a normal karyotype, and similar coexisting mutation profile including GATA2, WT1, EZH2, TET2, and NRAS mutations. Affected family members were observed to share almost identical comutations, indicating that germline CEBPA mutations may influence the acquisition and selection of specific cooperating mutations, which may be governed by the inherited genetic factors shared within families. FAML-CEBPA is associated with favorable long-term outcomes with a reported 10-year OS of 67%, but a higher relapse incidence (50-90% in different series). Multiple relapses may occur and can span decades. The first relapse usually presents later than sporadic cases. Median time to first relapse in FAML-CEBPA and sporadic ones were reported to be 27 months, and 11 months, respectively. Superior response to secondary therapies has been observed with a median survival of 8 years in contrast to 16 months in sporadic CEBPAdm following recurrence [18,45,65]. The genetic landscape and clinical heterogeneity of FAML-CEBPA are evolving with the more recent findings of cases with non-N-terminal germline mutations showing lower penetrance and longer latency. Long-term follow-up of these patients and symptomatic carriers would allow better characterization of the true prevalence and clinical and genetic heterogeneity of FAML-CEBPA and ultimately enhance patient care.

Laboratory Evaluation of CEBPA Mutations
With the wide breath of CEBPA mutations spanning from N-terminus to C-terminus and multiple mutation types (point mutations and indels), the most comprehensive method for CEBPA mutation detection is full length sequencing of this single exon gene. Before the wide clinical use of next-generation sequencing (NGS), it is commonly done by Sanger sequencing. As NGS has become routine in AML diagnostic work-up, CEBPA is frequently included in NGS panels which offer the convenience of simultaneous interrogation of multiple clinically informative genes. Sanger sequencing has a well-established analytical sensitivity of 15-20% whereas the sensitivity of many NGS panels is approximately 5%. Both platforms performing at the expected sensitivity are adequate to detect a germline mutation in the absence of events that may alter the variant allele fraction (VAF) such as concurrent copy number variation of the genetic region or allogenic stem cell transplantation. NGS can additionally detect somatic CEBAP mutations with lower VAF falling below the analytical sensitivity of Sanger sequencing. One technical caveat is that the CEBPA gene is GC-rich, necessitates vigorous validation and optimization during the test development phase to ensure reliable clinical performance. Capture-based NGS has been shown to perform better than amplicon-based NGS in CEBPA mutation detection [66]. As indel-type mutations are frequent in CEBPA, in medical facilities with limited resources for complex molecular testing, fragment analysis may be used as a screening tool to identify indel type mutations [3,67]. This method usually offers an analytic sensitivity around 5%; however, it will not detect point mutations or offer precise sequence information of the indel events. Furthermore, with the precision limitation of current capillary electrophoresis platforms, fragment analysis does not offer the capability of precisely determine the indel size (and therefore lack the ability to differentiate between in-frame and outof-frame indels) as the indel sizes increase. Historically, some labs also used denaturing high-performance liquid chromatography to screen for point mutations and small indels in CEBPA followed by sequencing in cases with abnormal chromatograms [11].
One important aspect of CEBPA testing is follow-up germline mutation confirmation in AML-CEBPA patients. This information is important for patient management including allogeneic stem cell transplant donor selection as well as patient family consultation and surveillance. In a diagnostic peripheral blood (PB) or bone marrow (BM) sample, typically a heterozygous germline variant in a treatment-naïve patient would show a VAF around 50%. However, VAF cannot be reliably used as a marker 1 3 to differentiate between germline versus somatic events in PB or BM as a VAF around 50% can also be seen in somatic settings and the VAF may be altered by other concurrent events such as copy number variation, loss of heterozygosity, and allogenic stem cell transplant. In AML patients undergo intensive chemotherapy, a strong clue for the presence of a germline CEBPA mutation is that at the time of complete remission, germline mutation persists (with a VAF around 50% similar to the diagnostic sample) while the other somatic mutation disappears [10,11,60]. For hematologic malignancies, PB or BM samples are not the appropriate constitutional sample type for germline testing in hematologic malignancies. The gold standard constitutional sample type for germline confirmation is cultured skin fibroblasts, with the caveat that the procedure is invasive and labor intensive. Other germline sample types include hair follicles, purified T-lymphocytes, saliva DNA, and buccal swab [9,10,18,68]. The rare occurrence of revertant somatic mosaicism seen in Fanconi anemia and dyskeratosis congenita has not be reported in FAML-CEBPA [69,70]. Family member testing with normal blood counts maybe be performed in PB DNA. However, these results should be verified with constitutional DNA, particularly in allogeneic stem cell transplant donor candidates. Germline genetic testing has profound psychosocial impacts, and the importance of multidisciplinary support and genetic counseling cannot be overstated.
Although biallelic mutation is required for AML with biallelic mutation of CEBPA, it is technically impossible for either Sanger sequencing or routine (non-long-range) NGS to definitively delineate whether the concurrent N-and C-terminal mutations are biallelic, as they are far apart on different amplicons (Sanger sequencing) or sequencing reads (by NGS), and may reflect true biallelic events occurring on separate alleles (in trans) in the same cell, or two mutations involving the same allele (in cis), or distinct occurrences in separate subclones. However, earlier cloning analysis has shown that the typical CEBPAdm pattern of concurrent Nand C-terminal/bZIP mutations is highly likely to represent a true biallelic mutation event [11,17]. In clinical practice, there are rarer occurrences of CEBPAdm not conforming to the typical CEBPAdm pattern with various combination of mutations occurring in N-, C-or middle regions of the gene. The clinical outcome of these cases has not been thoroughly studied in large cohorts, although few groups observed distinct methylation and gene expression patterns from the typical CEBPAdm and a trend toward worse OS particularly after relapse [40,71]. These findings add complexity and confusion to the precise assignment of "AML with biallelic mutation of CEBPA" in daily clinical practice.
Rather than the previous requirement for biallelic CEBPA abnormalities, the recent finding of favorable outcome of in-frame bZIP-mutation irrespective of double or single CEBPA-mutated status [12][13][14] has been adapted in both ICC and the 2022 ELN recommendation, while WHO-5 continues to include both CEBPAdm or bZIP CEBPA mutation [19,20,72]. In addition, ICC and ELN require a minimum of 10% blast for AML-CEBPA, whereas WHO-5 mandates 20% blast threshold. Hopefully, a consolidated definition of AML-CEBPA would be agreed upon among the WHO, ICC, and ELN in the near future to clarify the diagnosis and prognosis of this entity.

Treatment
In general, AML-CEBPAdm is a chemosensitive disease with a high response rate. The anthracycline and cytarabine based induction chemotherapy followed by consolidation remain the main therapy for this favorable risk AML in medically fit patients. Although the gemtuzumab ozogamicin has improved outcomes of cytogenetic favorable risk AML, its role in molecularly favorable AML like CEBPAdm remains unclear [73]. Recently, hypomethylating agents (HMA) coupled with venetoclax have been adapted as frontline therapy for patients who are not eligible for high intensity chemotherapy; however, favorable risk AML was excluded from the VIALE-A phase III randomized trial [74]. Arslan et al. studied the use of HMA + Venetoclax in favorable risk AML [75]. Thirteen (30%) out of 43 patients had AML-CEBPAdm. The complete remission (CR) and complete remission with incomplete count recovery (CRi) rate in this patient population were 75%.
MRD assessment is currently available for clinical practice. Deng et al. evaluated the role of multiparametric flow cytometry (MFC) MRD in CEBPAdm AML-CEBPAdm [76]. MRD-positive status during consolidation but not after induction was associated with an increased risk of relapse and decreased relapse free survival. However only elevated WBC count at time of diagnosis was prognostics of these outcomes in multivariate analysis. Wang et al. further stratified MRD as low risk and high risk MRD [77]. The low risk MRD was defined as negative MRD after at least two consolidation cycles of chemotherapy. Low risk MRD was not only associated with a lower risk of relapse, and improved relapse free survival (RFS) but was also associated with improved OS.
On the other hand, several early studies have identified the limited role of stem cell transplant in CR1 as it was not associated with improved OS [78]. Schlenk et al. studied the role of stem cell transplant in CR1 in the absence of MRD data. While both autologous and allogeneic stem cell transplant provide improved RFS in CR1, the OS was not different compared to patient who received chemotherapy only. This study provided evidence to reserve hematopoietic stem cell transplant to CR2 or refractory disease.
Relapsed AML-CEBPAdm continues to retain chemotherapy sensitivity. The second CR rate was reported to be 83-85% [78,79]. Wang et al. reported that patients who are in second CR or have refractory disease, benefited of allogeneic stem cell transplant [77]. Yet allogeneic stem cell transplant remains an important consolidative therapy in the prevention of future relapses in patient FAML-CEBPA, as it can treat the AML in addition to replace leukemia prone stem cells [53]. This highlights the importance of careful germline screening for patient with suspected FAML-CEBPA, as it would have implications on matched sibling donor selection since donor derived AML-CEBPA has been reported in the literature [80].

Conclusion
Here we reviewed the current knowledge of CEBPA mutation, co-mutation, and the recent advances and recognition of the favorable outcome of bZIP-mutation and its incorporation to the WHO-5, ICC, and ELN classifications. We also provided a detailed laboratory evaluation for CEBPA germline mutation for patient with suspected FAML-CEBPA. We elaborated on the current available therapies, the implications of MRD in stratifying the disease risk post therapy, and the role of stem cell transplant.
While AML-CEBPA generally harbors a favorable prognosis, further studies are required to address best strategies for the treatment of MRD-positive disease, and how to improve the outcomes of co-mutated GATA2, WT1, and CSF3R AML.

Conflict of Interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.