The Biology of Acute Promyelocytic Leukemia and Its Impact on Diagnosis and Treatment

Despite its rarity, acute promyelocytic leukemia (APL) is one of the most successful examples of translational research in medicine. In the past 15 years an extraordinary combination of laboratory and clinical research studies have in fact contributed to transform this once rapidly fatal disease into the most frequently curable adult leukemia.^1,2 Moreover, APL has represented a model to unravel key mechanisms of leukemogenesis and a paradigm for innovative treatments including differentiation therapy and the use of chromatin remodelling agents and antibody-directed therapy. In this review, the molecular and cellular features characterizing the disease will be reviewed, with particular emphasis on how these features impact on targeted therapy, diagnosis and disease monitoring (Figure 1; see Color Figures, page 514).

The genetic hallmark of APL is the balanced reciprocal translocation t(15;17)(q22;q11–12) leading to a fusion of the promyelocytic (PML) gene on chromosome 15 and the retinoic acid receptor-α (RARα) gene on chromosome 17. Depending on the location of breakpoints within the PML site, the PML/RARα transcript subtypes bcr1, bcr2 and bcr3 may be formed. Of these, bcr1 and bcr2 are of similar size and together referred to as long (L) isoform, bcr2 as variable (V) and bcr3 as short (S) isoform. The PML/RARα fusion is detectable by fluorescence in situ hybridization (FISH) or reverse-transcriptase polymerase chain reaction (RT-PCR) in > 95% of morphologically defined APLs, while in the remaining cases several variant rearrangements have been described that constantly involve RARα and partner genes other than PML.³ These alternative fusions may involve at low frequency (< 3%) the promyelocytic leukemia zinc finger (PLZF), or very rarely the nucleophosmin (NPM), nuclear mitotic apparatus (NUMA), and STAT5b partner genes. The nature of the fusion partner has an important bearing on disease biology particularly with respect to all-trans retinoic acid (ATRA) sensitivity, with APL due to involvement of PLZF being characterized by retinoid resistance.^3 –5 

RARα is a member of the RA nuclear receptor family that acts as ligand-inducible transcription factor by binding to specific response elements (RARE) at the promoter region of target genes. In the absence of ligand, RARα forms heterodimers with the retinoid X receptor (RXR) and recruits a co-repressor complex containing histone deacetylase (HDAC) activities that induces chromatin condensation and transcriptional repression. Physiological concentrations of RA (1 × 10⁻⁹ M) are able to release the nuclear co-repressors complex from the RAR-RXR and recruit co-activators with histone acetyltransferase activities (HAT). This results in hyperacetylation of histones at RARE sites, chromatin remodeling and transcriptional activation of RARα-target genes.^6,7 

PML belongs to a family of proteins containing a distinctive C₃HC₄ zinc-binding domain referred to as RING finger. Like other members of this family, including BRCA1, PML has been implicated in tumor suppression and control of genomic stability.⁸ PML controls p53-dependent induction of apoptosis, growth suppression, and cellular senescence in response to ionizing radiation and oncogenic transformation. Moreover, PML is required for transcriptional repression mediated by other tumor suppressors such as Rb and Mad.⁹ 

In the nucleus, PML is detected in multiprotein complexes termed PML nuclear bodies, where it co-localizes with other proteins such as p53, pRb, Daxx, and CBP.^8 –10 As a consequence of the PML/RARα fusion, in APL cells, PML is disrupted and the protein is de-localized into micro-speckled nuclear particles. This variation in subcellular distribution of PML in APLs bearing the PML/RARα rearrangement is clinically relevant as it allows rapid diagnosis through immunostaining techniques (see below).

The PML-RARα protein functions as an aberrant retinoid receptor with altered DNA-binding properties as compared to wild type RARα, and it is a repressor of RA signaling. PML-RARα can heterodimerize with the RXR, with PML and with another PML-RARα chimeric protein forming a homodimer. Distinct from wild-type RARα, the repression induced by unliganded PML-RARα is releasable only by pharmacological doses of ATRA (10⁻⁶ M). This is explained by the fact that, compared to wild-type RARα, the PML-RARα hybrid binds the histone-deacetylase (HDAC)–recruiting co-repressor complex with higher affinity.¹¹ In addition to recruiting HDAC activity, PML-RARα is also able to bind the DNA methylating enzymes Dnmt1 and Dnmt3a, leading to the methylation of RA target promoters in APL blasts.^12,13 In summary, PML/RARα acts through various mechanisms as a constitutive and potent transcriptional repressor of RARα-target genes.^11 –16 

In patients with the variant t(11;17) leading to the PLZF/RARα fusion, an additional co-repressor complex binding site is present in the PLZF moiety, accounting for stronger transcription repression and resistance to pharmacologic doses of RA. In these cases, histone deacetylase inhibitors have proven effective to restore sensitivity to retinoic acid in vitro.¹⁷ 

The variable types of PML/RARα isoform and their correlation with diagnostic features and outcome have been analyzed in several studies (Figure 2, see Color Figures, page 514).^2,18 These reports found an association of the S transcript with hyperleukocytosis, CD34 and CD2 expression, and M3v morphology. As to correlations with response to therapy, early studies in which ATRA was used as monotherapy reported less favorable outcome for patients with S-transcript compared to those with the L-isoform. Subsequent analyses on larger patient cohorts treated in the USA, Europe and Japan with modern ATRA plus chemotherapy regimens did not find significant differences in response to therapy according to the PML/RARα type, although a trend toward inferior outcome was reported in all studies for patients with the S transcript.² RT-PCR techniques used in these clinical studies, however, do not allow discrimination of bcr1 from bcr2 isoforms, as the latter analyzed as a single group. Therefore, the prognostic relevance of including the bcr2 patients in the L isoform group is unclear.

Alterations of the PML/RARα hybrid that may impact on therapy have been described to occur as acquired point mutations in about 30% of patients relapsing during RA therapy.¹⁹ These mutations, which are localized to the ligand-binding domain of the RARα region, may variably reduce RA binding and gene-regulatory activity. Interestingly, a recent study showed that clones harboring these alterations may develop in relapsing patients independently from proximate RA pressure.²⁰ 

In addition to representing a target for RA, the PML/RARα protein is also targeted by arsenic trioxide. This agent shows dual dose-dependent effects in APL cells including the induction of partial cell differentiation at low concentrations and of apoptosis at higher concentrations. Cell differentiation apparently occurs through PML/RARα degradation with consequent release of the maturation block. However, full differentiation to mature neutrophils does not occur with arsenic trioxide and differentiation is observed to a lesser extent as compared to that obtained with RA. At higher concentrations, arsenic trioxide inhibits APL cell growth in vitro, primarily through apoptosis that follows the induction of the proenzymes of caspase 2 and caspase 3, and activation of both caspase 1 and caspase 3.^21 –23 

Besides its unique genetic lesion, several other biological features contribute to the establishment of APL as a distinct entity within the acute myeloid leukemias. As discussed below, these features are clinically relevant because of their impact on disease presentation and targeted treatment, while their role in diagnosis and prognostic assessment is more controversial. They include a characteristic surface antigen profile, lack of the multidrug resistance phenotype and mutations in the FLT3 receptor (see below). Immunophenotypic studies have shown a distinctive and consistent antigen expression profile that includes strong positivity for CD33, expression of CD13 and CD117, infrequent expression of HLA-DR and CD34, and lack of CD7, CD11a CD11b, CD14 and CD18.^24,25 In addition, leukemic promyelocytes show low frequency of CD56 and CD65/CD65s expression, and differential reactivity with antibodies to CD15 and its sialylated form. Finally, the aberrant expression of the T-cell–associated antigen CD2 has been reported in a proportion of cases and associated with the microgranular (M3v) form and increased leukocyte counts at presentation.²⁶ Although low or negative in most cases, CD34 expression has also been reported in a subset of patients where it correlates with leukocytosis, hypogranular morphology and/or the S type of PML/RARα isoform.^25,26 A presumed negative impact on outcome has been suggested for some antigens (in particular CD34, CD2, CD56); however, studies in larger series of patients receiving homogeneous and state-of-the-art therapy have not confirmed these associations.^2,26 

Of the above antigens, CD33 is detectable in virtually 100% of cases and typically shows in APL a homogeneous expression pattern with increased antigen density. This may explain, at least in part, the good therapeutic response observed with either the conjugated or nonconjugated anti-CD33 antibodies which have been used in the clinic, i.e., gemtuzumab ozogamicin (GO) and HuM195, respectively. Both these antibody-based treatments proved highly effective as inducers of molecular remission even if used as single agents and/or in advanced disease.^27,28 

A further phenotypic characteristic of APL blasts is the absent or low expression of proteins associated to multidrug resistance (MDR) such as PGP, MRP1, MRP2, and LRP. In a series of APL patients analyzed at first presentation and at relapse, a study reported overexpression of PGP and LRP in 2% and 4% of cases, respectively, at the time of diagnosis and no changes in expression of MRP1 and MRP2. Interestingly, increased overexpression of only PGP and LRP was documented at relapse, while MRP1 and MRP2 remained unchanged.²⁹ Together, data on MDR appear relevant to explain the striking sensitivity of APL blasts to anthracyclines and GO.

The FLT3 gene encodes a tyrosine kinase III receptor involved in the proliferation and differentiation of hemopoietic stem cells. Upon binding to its ligand, FLT3 receptor activates multiple intracellular signalling pathways leading to cell proliferation and activation. FLT3 mutations including internal tandem duplications (ITD) in the juxtamembrane domain and point mutations in the tyrosine kinase II domain have been detected at high frequency in APL (up to 45% of cases). This finding is of particular interest given the opportunities for targeted therapies using FLT3 inhibitors. Both mutations have been consistently associated with higher white blood cell count at presentation, with ITD mutations being correlated with M3v subtype and S-type PML/RARα fusion.^30,–31 In addition, microarray analysis revealed differences in expression profiles between patients with mutated and wild-type FLT3, with patients in the former group showing increased expression of genes involved in cell growth, cell cycle control, cell adhesion and migration, and the coagulation/inflammation pathway.^31,32 Together, these studies suggest a role of FLT3 mutations in the pathogenesis and clinical manifestations of APL. Combined with the results of studies in transgenic mice for both PML/RARα and FLT3³³ these data also suggest that FLT3 mutations are most likely associated with disease progression. With respect to the prognostic significance of FLT3 alterations in APL, no significant correlation with response to therapy and survival were found in the two largest studies conducted so far that were reported by the GIMEMA and MRC groups.^30,31 Despite the promising results FLT3 inhibitors have shown in combination with RA in APL mouse models,³⁴ to date FLT3 inhibitors have not been evaluated in clinical trials in the treatment of patients with APL also owing to the availability of several highly effective agents already in use in this disease.

APL is a medical emergency frequently presenting with an abrupt onset. See the preceding chapter by Sanz. The high risk of early death (10–20%) and the potential for high cure rate (> 80%) account for the importance of immediate recognition and prompt initiation of specific treatment.

For the purpose of initial diagnosis, the morphologic appearance of hypergranular dysplastic promyelocytes allows typical cases to be identified and justifies immediate treatment initiation with RA, without waiting for diagnostic confirmation at the genetic level. Nevertheless, subsequent genetic diagnostic confirmation of APL is essential in all cases for several reasons: 1) it allows the clarification of cases with unusual morphology, such as the micro-granular M3v form, that are similarly responsive to RA and ATO; 2) it permits the exclusion of the rare variant translocations not involving PML that are resistant to RA; and 3) it allows for the identification of the precise target for disease monitoring through the characterization of PML/RARα isoforms by RT-PCR.³⁵ 

The identification of the APL-specific genetic lesion in leukemic cells is feasible at the chromosome, DNA, RNA, and protein levels with the use of conventional karyotyping, FISH, RT-PCR, and anti-PML monoclonal antibodies, respectively. Both RT-PCR and FISH have the additional advantage that no dividing cells are required for analysis, and they allow results to be obtained in cases with few or poor-quality metaphases, or where the PML-RARα fusion gene is formed as a result of cryptic or complex rearrangements in the absence of the classic t(15;17).³⁵ Because RT-PCR is notoriously prone to contamination and artifact, the assay should be carried out in experienced laboratories. Therefore, it is advisable that diagnostic and monitoring samples be sent to reference laboratories where well-trained personnel have experience on RT-PCR of PML/RARα. As to anti-PML immunostaining, this more recently introduced technique is very simple to perform and highly specific.³⁶ Using anti-PML antibody in indirect immunofluorescence or immunocytochemical assays, this method allows the differentiation of the microgranular nuclear distribution of PML associated with PML/RARα from the staining pattern referred to as “nuclear bodies” characteristic of other leukemias and normal hematopoietic cells. Results from the immunofluorescence assay can be achieved in only 2 hours. In light of its very convenient cost-benefit ratio, the assay is highly recommended to rapidly confirm diagnosis of APL at the protein level, also in small institutions not equipped and experienced for genetic analyses.^2,36 

The results of several longitudinal monitoring studies conducted in patients receiving modern RA plus chemotherapy regimens indicate that persistence of RT-PCR–detectable transcripts at the end of consolidation is strongly correlated with subsequent relapse and, conversely, repeatedly negative tests are associated with prolonged remission.^1,2,36 As a consequence of these observations, molecular remission is now established as a therapeutic objective in APL.³⁷ Two considerations are relevant to the interpretation of RT-PCR data and to optimize the clinical use of molecular monitoring using the conventional (non-quantitative) assay. First, the adequate time-point for assessing molecular response is the end of consolidation, i.e., following 3–4 cycles of RA and chemotherapy. Earlier assessment of response (i.e., after induction only) by qualitative RT-PCR, karyotyping or FISH may be misleading; various studies have demonstrated that evaluation at this time point is not predictive of outcome.² Second, the above reported correlations between molecular status and outcome have been established using low-sensitivity RT-PCR assays (with sensitivity threshold between 10⁻³ and 10⁻⁴), whereas methods with increased sensitivity have produced less informative data on association between molecular status at remission and prognostic outcome. Increasing RT-PCR assay sensitivity has led to occasional detection of PML-RARα transcripts in patients in long-term remission.³⁶ 

Molecular evaluation of PML/RARα has been more recently implemented by the introduction of quantitative PCR (RQ-PCR) tests that allow more precise and reproducible evaluation of leukemia-associated transcripts. Using this technique in patients in the US Intergroup APL trial, it was found that repeated testing during follow up significantly improves the ability to identify patients at higher risk of relapse.³⁸ In contrast, post-induction evaluation of PML/RARα transcript levels was poorly informative with respect to clinical outcome. In a recently published prospective study on 70 patients treated with RA and chemotherapy, it was shown that RQ-PCR allowed for identifying patients at high risk of relapse at an earlier time during treatment (post-first consolidation).³⁹ In another series of cases the levels of transcript assessed at 3 months post-therapy with RQ-PCR were predictive of successive outcome.⁴⁰ It is in general becoming increasingly clear that sequential monitoring by RQ-PCR following completion of consolidation provides the best approach to specifically identify the subgroup of patients destined to relapse. This technology is marginally more sensitive than conventional nested RT-PCR, and the majority of relapses can be successfully predicted by 3-monthly bone marrow assessment. This is in accordance with the maximal achievable sensitivity of RQ-PCR and the kinetics of relapse, whereby fusion transcripts typically rise by up to 1-log per month.⁴¹ 

Taken together, the results on RQ-PCR studies in APL suggest that this technology improves assessment of response in APL. However, to date this technique has been applied prospectively in only one clinical study.³⁹ The results in larger patient series such as those presently being analyzed by the US Intergroup and MRC are awaited to better establish the value of Q-RT-PCR in guiding the therapeutic decision. It is important to remember, finally, that the application of this method implies sample centralization, high level of expertise, and technologic facilities only available at present in more developed countries, while the correct use of qualitative RT-PCR evaluation may still represent a valid alternative in countries with limited resources.

Results of recent APL trials clearly indicate a benefit of using risk-adapted protocols in which patients are stratified according to clinical features at presentation. Monitoring of the molecular status during follow-up appears relevant in this context, as it may not only identify patients at risk who are in need of additional therapy but also spare excessive toxicity in patients with low risk of relapse. Moreover, studies of the GIMEMA group (reviewed in refs ^1,2) indicate that the administration of pre-emptive therapy in APL at time of molecular relapse provides a survival advantage over treating overt hematologic relapse.

Based on these considerations, most presently ongoing studies conceive longitudinal molecular monitoring of PML/RARα as an integral part of the treatment strategy and prospective PCR monitoring in the marrow is pre-established at fixed time points during follow-up. In these trials, patients defined as being at high risk of relapse include either those with elevated WBC count at diagnosis or those who persist or convert to PCR-positive for PML/RARα after consolidation and during follow-up. Accordingly, increased treatment intensity and/or early therapeutic intervention is pre-defined in the high-risk group as part of the study design.

In summary, RT-PCR may be intended as a relevant means to individualize treatment in this disease. It may allow the early identification of patients requiring additional therapy while minimizing the risk of over-treating patients who are likely to be cured with less intensive approaches. Modern therapeutic strategies exploiting the anti-leukemic effect of agents carrying decreased toxicity (RA, ATO, GO) are being used with greater confidence by investigators who rely on accurate molecular monitoring as a cautionary criterion to better direct treatment.

The body of available biological information on APL establishes this leukemia as a unique entity that has to be promptly recognized and clearly distinguished from all other acute leukemias, especially in light of its specific therapeutic requirements. As a model for targeted treatment, APL has fostered investigation of innovative therapies aimed at targeting molecular lesions and pathways related to neoplastic transformation. To date, however, the striking therapeutic progress achieved in APL has not been translated into other acute leukemia subsets, and attempts to release maturation block in these latter forms through differentiation therapy have been unsuccessful. The impact of laboratory evaluation on patient management and the rarity of the disease indicate that APL offers a unique opportunity for international collaboration between basic and clinical scientists.

Acknowledgments. We are indebted to David Grimwade for helpful comments, critical review of the manuscript, and providing unpublished data.