Childhood acute leukemias are frequent in Mexico City: descriptive epidemiology

Population-based, descriptive study.

For a case to be included in this study, the patient had to meet the following criteria:

All such cases from 2006 to 2007 were analyzed.

Although medical attention for children with AL is provided by different health institutions, both public and private, the public sector has been estimated to treat 97.5% of the cases of AL that occur in Mexico City [10]. Of the 13 hospitals that are equipped to treat children with AL in Mexico City, four were not included in this study because it had been determined in previous studies that these four hospitals (two hospitals of Petróleos Mexicanos, a Hospital Militar de México, and the Hospital Regional No. 25 of IMSS) had not treated cases of children from Mexico City. The nine hospitals that were included represent various governmental agencies: IMSS, SSa, ISSSTE, and the Secretaría de Salud del Distrito Federal (SSDF). These hospitals, in descending order of the proportion of cases each contributed to this study, were the following: Instituto Nacional de Pediatría (SSa), 27.6%; Hospital Infantil de México "Federico Gómez" (SSa), 20.2%; Hospital de Pediatría del Centro Médico Nacional "Siglo XXI" (IMSS), 19.3%; Hospital General "Gaudencio González Garza" of the Centro Médico Nacional "La Raza" (IMSS), 18.0%; Centro Médico Nacional "20 de Noviembre" (ISSSTE), 6.6%; Hospital Regional "Carlos McGregor Sánchez Navarro" (IMSS), 3.5%; Hospital Juárez de México (SSa), 2.6%; Hospital Pediátrico de Moctezuma (SSDF), 1.8%; and Hospital General de México (SSa), 0.4%.

In each participating hospital, we had a trained nurse or medical assistant for identifying cases of suspected acute leukemia. For each such case, after having signed a consent form, the parents were interviewed to determine demographic variables, and the patient's record was reviewed to obtain the clinical variables and the diagnosis. To ensure the quality of our data, the information was collected independently of the only existing Mexican Registry of Childhood Cancer (MRCC), which is maintained by IMSS and which contains data only for those served by IMSS [7]. The concordance between the data in our register for patients served by IMSS and the data in the MRCC was 100%.

Once diagnosed with presumed leukemia, a child was referred to one of the hospitals where trained staff (hematologists and onco-hematologists) did a comprehensive follow-up of the case to either confirm or discard the diagnosis of leukemia. Bone marrow smear was used to confirm each diagnosis; histochemical tests (myeloperoxidase, Sudan black B reaction, esterases, periodic acid Schiff (PAS) reaction, and acid phosphatase) were performed to differentiate the types of leukemia.

Morphological classification was used to divide the leukemias into five groups, according to the International Classification of Childhood Cancer (International Classification of Disease for Oncology) [14]. Only four of the five types were found in this study: a) ALL (9820-9827, 9850); b) AML (9840, 9841, 9861, 9864, 9866, 9867, 9891, 9894, 9910); c) CML (9863, 9868); and d) UL (9800-9804). The files were reviewed to corroborate that there were no duplicate records or inconsistent data. A database was generated to record age, sex, residency, year of diagnosis, and clinical manifestations of the patients.

Immunophenotyping was performed by flow cytometry. Cases of ALL were classified according to one of the following immunophenotypes: precursor-B cell, mature-B cell, T lineage, or "not otherwise specified", if the registered information did not allow proper classification. For the cases of ALL, the immunophenotypes registered were the following: for precursor B-cell ALL, CD19+, HLADR+, cyCD22+, CD10+ or CD10, TdT+, CD20+, cyCD79a+, and CD34+; for T lineage ALL, CD19, CD22, CD79a, CD7+, CD5+, cyCD3+, clgm, and sIg; and for mature B cells, immunoglobulin (kappa or lambda light chain as surface markers).

Only two risk categories were used for this study: 1) children, aged 1-9 years, with a white blood cell count less than 50,000/μL were classified as being at standard risk; 2) children either who were in the 10-14 age group or who were younger and had a white blood cell count greater than 50,000/μL were designated as being at high risk [15].

Because the size of the base population, estimated from the data for Mexico City from the Instituto Nacional de Estadística, Geografía e Informática (INEGI) [16], was known, it was feasible to obtain the incidence rate for the population under 15 years of age (Additional file 1 Table S1). The denominator was calculated by using data from the intermediate census of 2005. In 2005, the Distrito Federal had a population of 8.72 million inhabitants, 2.04 million of whom were children under the age of 15 years [16]. With no census having been carried out in 2006 or 2007, there is no official count for those years; therefore, as estimation for those two years, the 2005 value for the population under 15 years of age was multiplied by two. (The population of Mexico City has remained quite stable from 1990 to date, with changes of only 0.04% per year [16].)

AAIR were calculated in total, by kind of leukemia, by age group (< 1 year, 1-4 years, 5-9 years, or 10-14 years), and by sex. The standardized average annual incidence rates (SAAIR) were standardized by age by using the direct method with the world standard population [17], reported per million.

To determine whether the SAAIR of leukemias varied by some characteristic in the boroughs of Mexico City, the correlation, as calculated by using Pearson's r, between the incidence of ALL and that of AML and the MHDI [18] for Mexico was determined. The value of P was reported, with P = 0.05 used as the cutoff value for statistical significance. The MHDI is a measurement of how well cities and their subdivisions are doing. The values of this index range between 0 and 1, with values closer to 1 signifying a greater degree of well-being.

Three indicators are used to construct this index:

Information concerning this index was available only for the year 2005 [16]. Each borough of Mexico City was analyzed both according to the hectares used for agriculture contained within its boundaries and according to the average number of persons per household, information that also was available only for the year 2005, [16] (data not shown).

This study was approved by the Ethics Board of the National Commission of Scientific Investigation (Registry No. 2008-785-063).

The overall SAAIR was 57.6 per million (95% CI, 46.9-68.3), with the SAAIR for ALL being 49.5 per million; for AML, 6.8 per million; and for CML, 0.9 per million. The male:female ratio was 1.0 for leukemias in general, 1.1 for ALL, 0.9 for AML, and 1.0 for CML.

The AAIR for ALL was highest (77.7 per million) for the 1-4 year age group. In this age group, the AAIR for ALL for males was higher than that for females (87.1 and 67.8 per million, respectively), whereas for the 5-9 year age group, the AAIR for ALL was higher for females (Table 1).

The immunophenotypes were determined for 96.4% of the ALL cases: 73.2% of the total ALL cases were classified as precursor B-cell; 12.4% as T cell; 8.2% as B cell; and 2.1% as dual phenotype, with 0.5% as undeterminable. The SAAIR of the pre-B ALL was 35.8 per million and that of the T-cell ALL was 6.3 per million (Table 2).

Of the ALL patients, 58.8% were classified as being high risk; when only those cases of ALL having precursor B- cell immunophenotype were considered, 53.5% were high risk.

Of the cases of AML, the most frequent classifications were M2 with seven cases; M4, with six; and M1 and M5, with five cases each. M3 represented only 10.7% of the cases.

As determined from the graph (Figure 1) the peak age for AL was found to be between 2-6 years of age for ALL, with another noteworthy peak at 8-10 years of age. These age peaks corresponded to those for precursor B-cell ALL. For T-cell ALL, a small peak was seen between 1-4 years of age. The AML showed a peak at one year and between 10-11 years (Figure 1).

The SAAIRs for the boroughs of Mexico City ranged from 23.0 to 87.7 per million, with Cuauhtémoc, a relatively affluent borough, having the highest SAAIR (Table 3).

The MHDI of the boroughs of Mexico City (Figure 2A, B) showed a negative correlation both with the incidence of ALL (Pearson's r, -0.138; P = 0.30) and with the incidence of precursor B-cell ALL (Pearson's r, -0.185; P = 0.49); in both instances, there was little precision in the estimation. No correlation was found between the number of cultivated hectares and the incidence of either AML, ALL, or precursor B-cell immunophenotype. However, there was a statistically significant, positive correlation between the average number of persons per household and the incidence of the pre-B immunophenotype (Pearson's r, 0.789, P = 0.02).

For Mexico City, the frequency of leukemia is higher than those for other cities. For cities in Canada, the USA, or the UK, the SAAIRs are 50.8, 46.9, and 40.8 per million, respectively [21], whereas the SAAIR for Mexico City was 57.6 per million children, as determined in this study. There are several factors that could affect the results: 1) Length of study period. Due to funding constraints, we were able to carry out our study only over a two-year period; therefore, the representativeness of the sampling could be questioned. However, the SAAIR is similar to those published by IMSS for the years 1996-2000 [6] and 1996-2002 [7] (58.4 and 55.4 per million children, respectively). 2) Value used in the denominator. Although we used 2005 data (then the most current official information) for the denominator in the calculations for 2006 and 2007, we are confident that little error was introduced, because over the last 20 years, the population of Mexico City has remained stable [16]. 3) Scope of sampling. This report does not contain information about the children treated in private institutions. However, because in Mexico City, nearly 97.5% of children with leukemia receive medical care in public institutions [12], the exclusion of this small percentage of patients in private institutions should not affect the main conclusions of the current study. In fact, inclusion of said cases would result in a larger numerator, thus leading to an even higher incidence rate. For these reasons, we think that the above-mentioned considerations did not affect the results. We had the opportunity to corroborate only the concordance between the cases of IMSS patients, which were registered in this study, and the data from the MRCC. For the data obtained from the other hospitals, there was no other registry with which to corroborate that the information was complete. Nevertheless, for this study, the registering of cases was done actively, in that a nurse visited the participating hospitals daily in order to identify any child diagnosed with suspected leukemia. Once so diagnosed, the patient was followed until the diagnosis was confirmed. The information in this study was compared to the admissions lists of the hospitals in order to verify that no patient, who had been admitted with the diagnosis of leukemia, had been overlooked by the nurses that performed the data collection. Another factor that gives us confidence in the numerator obtained in this study is that the SAAIR of leukemias (57.6 per million) is similar to those reported in studies by the IMSS (55.4 and 58.4 per million [6, 7]). The especially high incidences, reported for children from Mexico City [6] and Costa Rica [5]; for Hispanic children in Florida [22], Los Angeles [23], and Texas [24] in the USA; and for Hispanic children in general as reported by the SEER and the CDC (Table 4) [1‐3, 25, 26], are due to the incidences of lymphoid leukemia in these populations. This finding is very interesting because, according to data from El Salvador [6], Argentina [27], and Brazil [28] and according to the International Report of Cancer in Children, populations in other regions of Latin America do not have higher incidences of lymphoid leukemia than do populations of Caucasian origin [21].

The high incidence in some Hispanic groups may have resulted from an artifact in the data, or have been due to environmental or genetic factors. In previous reports, there may have been an artifact in the IMSS data, which resulted in an overestimation of the incidence. The following scenario might explain such a situation: unemployed parents, on learning that their child had leukemia, would seek employment that would provide access to medical benefits from IMSS. The child would then be included in the numerator, without being represented in the denominator, thereby leading to an overestimation of the incidence. (A similar situation could be envisioned for the Hispanic population in the USA, many of which are undocumented workers. Illegal aliens may be forced out of hiding to seek medical attention for their child, thereby increasing the numerator). It should be noted that the high incidence demonstrated in the current study agrees with those found in prior reports based on IMSS data, implying that the prior reports were not biased, despite having come from only one institution.

Another possible explanation of the higher incidence among some groups of Hispanics is that exposure to carcinogenic agents is greater among Hispanics in the USA than it is for other ethnic groups [29], as a significant portion of Hispanic immigrants in the USA are employed in high-risk occupations, such as agriculture in which pesticides are used. This is especially true for Florida, California, and Texas [29], the states in which higher incidences of ALL among Hispanic children have been reported [22‐25]. We have previously reported that, in Mexico City, both exposure to carcinogenic agents in the workplace and agriculture-related occupations were risk factors for AL in children [30]. However, in this study, for the boroughs of Mexico City, no correlation was found between agricultural hectares and the incidence of ALL. Another possibility is that genetic factors could explain why Costa Ricans, Mexicans, and Hispanics in the USA have the highest incidences of ALL [23]. In contrast, the SAAIR of AML for the study population was 6.8 per million, a value similar to data from Canada, the USA, and the UK (SAAIRs of 6.3, 6.0, 6.3 per million, respectively) [21].

First, it should be noted that, when the age groups for ALL are compared across various groups, the incidence for children 1-4 years of age were consistently greater than those for 10-14 year olds: for the majority of Mexican states, two-fold greater (a finding in agreement with the IMSS data); for the state of Nuevo León, three-fold; and for developed countries, more than three-fold, even though the rate of ALL is not so high [7]. In the current study, the incidence for 1-4 year olds was 2.6-fold that for the 10-14 year olds. This relation has been associated with socio-economic level and with the possibility that a hypothetical infectious agent may be involved [13]. It is interesting that the frequency of ALL for Nuevo León, a Mexican state that borders the USA, is similar to that for the Caucasian population of the USA [7], the fact that Nuevo León has the highest MHDI of Mexico, suggests the possibility that, among Hispanic children who enjoy a higher quality of life, the incidence of ALL tends to follow that of developed countries. Further studies are required to validate this hypothesis.

In this work, the frequency of immunophenotyping was 96.4%, a value higher than those in other published data from Mexico [8] but similar to those in reports from developed countries [31]. The frequency of precursor B-cell ALL was similar to those in other reports from Mexico [8, 32, 33]; the frequency of T-cell ALL was lower than that (23.6%) in a previous report from IMSS [30], but similar to those in the latest reports from the Instituto Nacional de Pediatría [32, 33].

The higher incidence of ALL among children over ten years of age is important because all children of this age are generally considered at high risk for relapse [31, 34]. In Mexico City, the ratio of high-risk children to standard-risk children is 1:1, a value in sharp contrast with those for populations attended in hospitals outside Mexico, e.g., St. Jude Children's Research Hospital (Memphis, Tennessee, USA) where the ratio is 1:3. Thus, only 25% of the children at Saint Jude [35] are at high risk, compared to 50% in Mexico City [36].

The importance of a high incidence of ALL at an early age is that it is a predictor of a higher frequency of a genetic rearrangement that has a good prognosis for ALL patients [37]. In Mexico, genetic rearrangements (such as ETV6/RUNX1) that have a good prognosis have been found at a frequency below those of developed countries [38], whereas genetic rearrangements with a bad prognosis have been reported at a higher frequency [38]. Of note is the high frequency found for MLL/AF4 in Mexico City, because this rearrangement, which has a very bad prognosis, has been related with different intrauterine exposures [39].

For Hispanic populations, AML M3 has been proposed as the predominate subtype [7, 40]. This idea was not corroborated in the present work: AML M3, with a frequency of only 10.7%, was not the predominant subtype in the population studied.

In a previous study in Mexico City, peaks of incidence of ALL were at 2-3 and at 6-8 years of age [8]; here, as shown in Figure 1, two peaks were found, but at 1-6 years of age and at 9-10 years of age. In other Hispanic populations in which the incidence of ALL is high [41], a similar situation seems to exist, that of there being two age peaks, one early, one late. A small peak for T-cell ALL occurred at 1-4 years, similar to that published by the IMSS from Mexico [8].

In this study, there was a strong correlation between the incidence of precursor B-cell immunophenotype and the average number of people per household in the boroughs of the city. This finding supports the infectious agent hypothesis, because a child living a crowded household would have a higher risk of being in contact with infectious agents [13]. In our report, the incidence of AML was not correlated with the agriculture hectares, average number of people per household, or MHDI. The incidence in Mexico City is similar to that of other cities in the region [21].