Cervical cancer (CC) is the second cause of death by cancer among women in the world and the first in most developing countries [1, 2]. In the year 2000, 190,000 deaths and 80% of the 500,000 new CC cases occurred in the developing world, and Latin American countries were among those with the highest incidence rates, together with countries from Sub-Saharan Africa, South and South East Asia [3]. The risk of developing CC increases with early start of sexual activity, number of sexual partners, prolonged use of oral contraceptives and smoking [4]. In Mexico it is associated with poverty related factors such as low schooling, unemployment, residence in rural areas and lack of access to health services [5–7].

zur Hausen [8] proposed and later demonstrated [9] that human papillomaviruses (HPV) are the infectious agents responsible of the neoplastic transformation of the cervical epithelium. This hypothesis was validated by finding HPV genome sequences in 99.7% of invasive CC cases [10]. It is currently accepted that HPV is the most common sexually transmitted pathogen [11] and that infection of the cervix by 'high risk' (HR) HPV types is a necessary factor for CC development [10, 12]. Low risk (LR) HPV types are those usually found in warts and benign lesions whereas HR types are those found in invasive CC [13].

There are 11 major LR-HPV types (designated as 6, 11, 40, 42, 43, 44, 54, 61, 70, 72, 81) and 15 major HR-HPV types (designated as 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, 73, 82) [14]. The two most frequently associated to malignancies are HR-HPV types 16 and 18. The first is responsible for nearly 50% and the second for nearly 20% of all invasive CC cases in the world [10, 15]. Several HR-HPV variants with higher oncogenic potential are more prevalent in developing countries, where they appear to contribute to the higher incidence and mortality rates. In Mexico the probability of developing CC by the Asian-American variant of HPV type 16 is several times higher than by the European variant, and around 25% of invasive CC cases are attributed to the AA variant [16].

During the year 2000 the incidence of CC in Mexico was 40.5 cases per 100,000 women over 25 years of age and the mortality rate was 17.1 [3], which means 12 deaths per day. In the Mexican states of San Luis Potosí (SLP) and Guanajuato (Gto) the mortality rates for the same year were respectively 15.5 and 21.7 [17]. Lazcano-Ponce et al. [18] found a HPV prevalence of 16.7% in Mexican women below 25 years of age which decreased to 3.7% in those of 25–44 years and increased to 23% in those above 65 years, and that HR-HPV types are more prevalent in all age groups whereas LR-HPV types are less prevalent in women under 25 years and increase with age.

In the present work we determined the prevalence of the HR-HPV types circulating in the neighboring states of San Luis Potosí and Guanajuato by restriction fragment length polymorphism (RFLP) analysis with the method of Fujinaga et al. [19], in women with precancerous and cancerous lesions of the cervix that had been diagnosed by cytological analysis; in two thirds of them biopsies or surgical specimens were obtained and the microscopic diagnosis confirmed.

Studies on the prevalence of HPV cervical infection in Mexico are scarce [7, 20–22] and information on HR-HPV prevalence is even scarcer [12, 22] or deals only with HPV types 16 and 18 [23, 24].

Since our aim was to determine the prevalence of the HR-HPV types causing cervical infections in the states of SLP and Gto, the study included women with cervical intraepithelial lesions (73.1% low grade, 16.7% high grade and 10.2% invasive carcinoma), expected to be HPV-positive. This prediction was fulfilled because nearly all women were shown to be infected with HPV, and HR types were identified in most of them.

To determine HPV infection and to identify the high-risk types involved we used the PCR-RFLP method based on amplification of the E6-250 HR-HPV product followed by digestion with endonucleases [19] and fragment sizing in high resolution agarose gels [25]. Obtaining cervical scrapings with the cytobrush allowed us to obtain more than enough amplifiable DNA in nearly all samples and PCR optimization assured efficient amplification with the primer sets employed.

The E6-250 HR-HPV product was amplified through nested PCR because this modality turned out to be at least twice as sensitive as direct PCR. The universal primer sets for the first ('direct') reaction amplified the E6-650 product [19, 26] and the primer set used in the second ('nested') reaction amplified the E6-250 product [27]. The use of duplex PCR was discarded to amplify E6-250 and L1-450 due to the interference observed when simultaneous amplification of both products was attempted (Fig. 4). The scarce E6-negative samples were tested later with a universal primer set that amplifies L1-450 segments of both low risk- as well as HR-HPV types [28]. From these observations we concluded that to determine HR-HPV infection in women with dysplasia and neoplasia it is convenient first to identify the HPV-positive samples and then to check the amplificability of the negative ones.

Eight HR-HPV types (16, 18, 31, 33, 35, 45, 52 and 58) are responsible for 95% of the invasive CC cases in the world [14]. The PCR-RFLP used in this work identified all of them except type 45.

Considering all cervical samples included in this work, 98.8% contained amplifiable DNA, 97.8% had HPV infections which in 90.2% of the cases corresponded to HR-HPV types whose prevalence was higher in Gto (99.0%) than in SLP (82.6%) (Table 3).

In the 437 samples with amplifiable DNA included in this study, 528 cervical infections were demonstrated by HR-HPV types that were identified in 99.1% of the cases (Table 4). In this way we demonstrated that in our region circulate the seven HR-HPV types identified with the method used and their overall prevalence is, in descending order, 16 (53.4%) > 31 (15.6%) > 18 (8.9%) > 35 (5.6) > 52 (5.4%) > 33 (1.2%) > 58 (0.7%) = unidentified types (0.7%). From these results we recommend that for analysis of cervical samples with dysplasia it is convenient to identify first the types 16, 18 and 33 by restriction with Ava II and then the types 31, 35, 52 and 58 by restriction with Rsa I, Ava I, Bgl II and Acc I.

The prevalence of high and low risk HPV types found by us is similar to that already seen in Mexico and other regions of high CC incidence [20, 21, 29]. The predominant HR-HPV types in our region are 16, 31 and 18 which together represent 77.9% of all cervical infections. The four remaining types cause 12.9%, and unidentified types only 0.7% of the HR-HPV infections. Types 16 and 18 are slightly more prevalent in SLP, whereas type 31 is nearly two times more prevalent in Gto.

The overall prevalence of HPV type 16 in our region (53.4%) is almost identical to the world average (53%) calculated by Munoz (2000). In contrast, the regional prevalence of type 18 (8.9%) is substantially lower than the world average (15%) calculated by Munoz [21]. The prevalence of types 31 (15.6%) and 35 (5.6%) is higher than the average for Latin America [21], whereas that of type 33 (1.2%) is nearly 10 times lower than the average for Central America (11.8%) recently obtained by Clifford et al. [30].

Single and triple infections were more frequent in SLP, whereas double infections were more frequent in Gto. Single infections included all seven HR-HPV identifiable types where the three predominant ones were, in descending order, 16 > 18 > 31 with a cumulative frequency of 92.7%. The prevalence of these types in SLP was, in descending order, 16 > 18 > 31 with a cumulative frequency of 95.0%, whereas in Gto it was 16 > 31 > 18 with a cumulative frequency of 89.1%.

The typing method allowed us to identify a considerable proportion of infections by two viral types (overall prevalence of 25.9%, with 22.6% for SLP and 29.5% for Gto) and three viral types (overall prevalence of 2.7%, with 2.4% for SLP and 3.2% for Gto) (Table 4). Type 16 was found in all multiple infections.

Double infections from SLP were caused by the seven identifiable types but type 58 was absent in Gto. The overall predominant types in double infections were 16 > 31 > 35 > 18 = 52. Their prevalence differed in each state since the descending order in SLP was 16 > 31 > 18 > 35 > 52, whereas in Gto it was 16 > 31 > 35 > 52 > 18 (Table 6).

Types 33 and 58 were absent from triple infections in both states and the overall predominant types were 16 > 35 > 31 = 52 > 18. There were also differences in the prevalence by state since in SLP the descending order was 16 > 35 = 52 > 31 > 18 whereas in Gto it was 16 > 35 = 52 > 31 > 18 (Table 6).

Our algorithm for typing, designed to identify first the high-risk viral types and then the low risk ones, probably underestimates the prevalence of low risk types in samples where high risk types are demonstrated. However, it is worth noting that in low grade intraepithelial cervical lesions practically all genital HPV types are found, whereas in high grade lesions only high risk types are found [31].

The proportion of multiple infections found by us (28.6%) is over three times higher than the world average (8.1%) calculated by Molano et al. [32]. This feature is probably due to the fact that the women in our study are themselves a high risk group since they were included for having cervical dysplasia and cancer. This idea is reinforced by our finding that 10.6% of the double infections are due to the 16/18 type couple, whose prevalence was nearly six times higher than the world average (1.8%) estimated by Clifford et al. [30].

The five predominant HR-HPV types (16, 31, 18, 35 and 52) found in the 140 HSIL and invasive cancer lesions accounted for 92.9% of the cases and their frequency was similar to their overall prevalence (Table 9).

Infections by two or more HPV types are markers of persistent cervical disease, of multiple cervical lesions and of progression from low to high grade lesions [33]. There is scarce evidence on the interaction of HPV types to enhance CC pathogenesis and it is unknown if some HPV types promote or exclude infection by others. Careful follow up of multiple infections may help to pinpoint the relevance of the interactions of different HPV types in disease progression [34].

HPV16, the most frequent type in single and multiple infections in SLP and Gto, just as in the rest of the world [14, 21] appears to be the most persistent type [32, 35, 36] increasing the risk of developing severe preneoplastic lesions [37].

Our work confirms the usefulness of the PCR-RFLP method of Fujinaga et al. [19, 26, 27] for diagnosis and typing cervical infections by HR-HPV, whose sensitivity was at least doubled by amplifying the E6-250 product through nested rather than direct PCR and whose typing precision was improved by using high resolution electrophoresis. In this way we could identify the HR-HPV types circulating in our region and their distribution in the two neighboring states analyzed, which reflect the endemic behavior of the various HPV types as well as local environmental or genetic conditions of the hosts [38].

Besides being of use to carry out molecular epidemiology studies and to predict the individual risk of each patient, HR-HPV typing is essential to plan the administration of prophylactic vaccines directed against one or more specific HPV types, which is questionable in populations where prevalence of the circulating types is unknown [31, 39, 40]. Since in our region HPV type 16 has the highest overall prevalence (53.4%), type 31 is second (15.6%) and type 18 is third (8.9%), the vaccine against HR-HPV types 16 and 18 [41] whose license was granted first in Mexico and the United States [42, 43] would be expected to cover at least these three types [44] and would have a 78% efficacy.

Enough DNA is obtained from cervical scrapings of women with dysplasia in order to detect high risk-HPV by PCR amplification followed by typing through restriction fragment length polymorphism analysis.

Sensitivity of HR-HPV detection is nearly doubled by nested rather than direct PCR amplification.

Ninety-two percent of the 437 Mexican women with cervical dysplasia and cancer included in this study were infected by HR-HPV, with a prevalence 1.14 times higher in the state of Guanajuato than in San Luis Potosí.

All seven HR-HPV types identifiable with the method used (16, 18, 31, 33, 35, 52, 58) circulate in SLP and Gto and were diagnosed in nearly all (99.3%) the cases.

Seventy-one percent of the infections were caused by one HR-HPV type, 25.9% by two types and 2.7% by three types.

The overall frequency of cervical lesions was LSIL (73.1%) > HSIL (16.7%) > invasive cancer (10.2%) and the ratio of single to multiple infections was inversely proportional to the severity of the lesions: 2.46 for LSIL, 2.37 for HSIL and 2.15 for invasive cancer.

The frequency of HR-HPV types found in HSIL and invasive cancer was 16 (55.0%) > 31 (18.6%) > 35 (7.9%) > 52 (7.1%) > 18 (4.3%) > unidentified types (3.6%) > 33 (2.9%) > 58 (0.7%).

Since the three predominant HR-HPV types found (16, 31 and 18) cause 77.9% of the HR-HPV infections associated to cervical dysplasia in this region, and immunization against type 16 prevents type 31 infections, in this region the efficacy of the prophylactic vaccine against types 16 and 18 would be close to 80%.